Logseq vs Reflect

Agent access

1. ai-native userPoint an agent at llms.txt or agent-oriented docs

   weight 2 · round drawn
   Logseq—none0/10

   There is no llms.txt file (probe returned 404) and no evidence of agent-oriented docs formatted for LLM consumption; the docs are standard human-facing markdown/wiki pages.

   • [probe] “PROBE llms.txt: HTTP 404 at https://plugins-doc.logseq.com/llms.txt”
   • [probe] “PROBE openapi: all candidate paths 404 (https://plugins-doc.logseq.com/openapi.json, https://plugins-doc.logseq.com/swagger.json, https://pl…”

   proof ↗⚑ flag

   Reflect—none0/10

   There's no llms.txt (404 confirmed by probe) and no agent-oriented documentation aimed at AI agents; Reflect's docs are human-facing product docs, not machine-consumable agent instructions.

   • [probe] “PROBE llms.txt: HTTP 404 at https://reflect.academy/llms.txt”

   proof ↗⚑ flag

2. ai-native userRun the product headlessly / in CI for automation

   weight 2 · round drawn
   Logseq—none0/10

   Logseq is a desktop/mobile Electron app with a local HTTP API and plugin system, but nothing in the evidence describes a headless mode, CLI, or CI-runnable binary for automation — the HTTP API requires the desktop app to be running interactively. Missing for 10: any documented headless/CLI mode, CI examples, or automation-oriented server mode.

   • [claimed-docs] “local HTTP server to access logseq's graph ... Enable HTTP APIs server in `Features > HTTP APIs server`”
   • [claimed-docs] “we developed the local HTTP server to support external access to the database through authorization.”
   • [claimed-docs] “Logseq: A privacy-first, open-source knowledge base”

   proof ↗⚑ flag

   Reflect—none0/10

   The axis applies to this product kind (peer products hold positive or none verdicts on this story), so lack of evidence for an applicable capability is "none", never "na". (na/none harmonized at arena bring-up — see pipeline/scripts/na-harmonize.ts.)

   ⚑ flag

3. ai-native userPlug MCP servers into this product so it can use their tools

   weight 3 · round drawn
   Logseq—none0/10

   Logseq documents a plugin system, HTTP API, and URL protocol for third-party integration, but nothing in the evidence pack mentions MCP servers or any mechanism for Logseq to consume MCP tools. Community comments even lament that AI tools like Claude/Codex can't be used to update pages, reinforcing the absence of MCP support.

   • [claimed-docs] “Extends the functionality of logseq with third party tools ... third-party plugins from the Marketplace, can be searched, installed, updated…”
   • [claimed-docs] “local HTTP server to access logseq's graph ... Enable HTTP APIs server in `Features > HTTP APIs server`”
   • [claimed-docs] “we developed the local HTTP server to support external access to the database through authorization.”
   • [community] “I've got so much already in logseq though, and I really like the model. Right now I'm thinking I'll just stay on version one as long as poss…”

   proof ↗⚑ flag

   Reflect—none0/10

   The axis applies to this product kind (peer products hold positive or none verdicts on this story), so lack of evidence for an applicable capability is "none", never "na". (na/none harmonized at arena bring-up — see pipeline/scripts/na-harmonize.ts.)

   ⚑ flag

4. ai-native userConnect an agent via an official MCP server

   weight 3 · round drawn
   Logseq—none0/10

   Logseq documents a local HTTP API and a plugin ecosystem for third-party integrations, but there is no mention anywhere in the evidence of an official Model Context Protocol (MCP) server or endpoint. Since Logseq is a knowledge-base platform (not itself an AI agent), shipping an official MCP server is a fair axis to ask about, but no such capability is evidenced.

   • [claimed-docs] “local HTTP server to access logseq's graph ... Enable HTTP APIs server in `Features > HTTP APIs server`”
   • [claimed-docs] “we developed the local HTTP server to support external access to the database through authorization.”
   • [claimed-docs] “Extends the functionality of logseq with third party tools ... third-party plugins from the Marketplace, can be searched, installed, updated…”

   proof ↗⚑ flag

   Reflect—none0/10

   The axis applies to this product kind (peer products hold positive or none verdicts on this story), so lack of evidence for an applicable capability is "none", never "na". (na/none harmonized at arena bring-up — see pipeline/scripts/na-harmonize.ts.)

   ⚑ flag

5. ai-native userUse an official CLI

   weight 2 · round drawn
   Logseq—none0/10

   No evidence of an official Logseq CLI; the product exposes a local HTTP API, a URL protocol scheme, and plugins, but nothing about a command-line interface for AI-native automation. Missing for 10: any mention of an official CLI tool, its command set, or installation/usage docs.

   ⚑ flag

   Reflect—none0/10

   The axis applies to this product kind (peer products hold positive or none verdicts on this story), so lack of evidence for an applicable capability is "none", never "na". (na/none harmonized at arena bring-up — see pipeline/scripts/na-harmonize.ts.)

   ⚑ flag

6. ai-native userDrive the product through a documented public API

   weight 3 · round drawn
   Logseq~partialprobed6/10

   Logseq documents a local HTTP APIs server explicitly for 'external access to the database through authorization' and a separate community-maintained plugin API reference, plus a logseq:// protocol for third-party invocation — all documented, public interfaces for programmatic control. However, there's no formal machine-readable spec (probes show 404s for openapi.json/llms.txt) and a community report explicitly states losing the ability to use Claude/codex to write/update pages after a version change, indicating real-world friction for AI-driven API use. Missing for 10: a formal OpenAPI/machine spec, first-party confirmation of AI-agent compatibility, and independent corroboration that the HTTP API reliably supports agentic write workflows.

   • [claimed-docs] “local HTTP server to access logseq's graph ... Enable HTTP APIs server in `Features > HTTP APIs server`”
   • [claimed-docs] “we developed the local HTTP server to support external access to the database through authorization.”
   • [claimed-docs] “This is a community-driven documentation for Logseq plugin APIs.”
   • [claimed-docs] “Logseq provides a protocol `logseq://` to allow third party applications to invoke actions within Logseq.”
   • [probe] “PROBE llms.txt: HTTP 404 at https://plugins-doc.logseq.com/llms.txt”
   • [probe] “PROBE openapi: all candidate paths 404 (https://plugins-doc.logseq.com/openapi.json, https://plugins-doc.logseq.com/swagger.json, https://pl…”
   • [community] “I've got so much already in logseq though, and I really like the model. Right now I'm thinking I'll just stay on version one as long as poss…”

   proof ↗⚑ flag

   Reflect~partialprobed6/10

   Reflect documents a public REST API with OAuth2 (including PKCE) authentication for building third-party clients, and a runtime probe confirms the API endpoint is live and enforces auth as documented. However the API is explicitly limited to appending data and listing bookmarked links (not full CRUD due to E2E encryption), and there is no formal OpenAPI/swagger spec or llms.txt discoverability (both probed and returned 404s). Missing for 10: comprehensive OpenAPI/swagger schema, broader read/write API coverage beyond append-only actions, official SDKs, and machine-readable API discovery file.

   • [claimed-docs] “For authentication, we support OAuth 2. You will need to create credentials here.”
   • [claimed-docs] “Our supports appending data to your notes, returning a list of the links you've bookmarked, and a few other things.”
   • [claimed-docs] “We do support OAuth's PKCE extension which means you can build a client to Reflect without shipping secret keys around.”
   • [probe] “PROBE runtime (recorded 2026-09-05): keyless GET to the documented REST base https://reflect.app/api/graphs returned HTTP 401 {"error":{"typ…”
   • [probe] “PROBE llms.txt: HTTP 404 at https://reflect.academy/llms.txt”
   • [probe] “PROBE openapi: all candidate paths 404 (https://reflect.academy/openapi.json, https://reflect.academy/swagger.json, https://reflect.academy/…”

   proof ↗⚑ flag

7. ai-native userIssue scoped/least-privilege API credentials for an agent

   weight 2 · round drawn
   Logseq—none0/10

   Logseq's local HTTP API server requires a single authorization token for full graph access, but there is no evidence of scoped or least-privilege credential issuance (e.g., read-only vs write, per-agent scopes, or permission granularity) for AI agents.

   • [claimed-docs] “local HTTP server to access logseq's graph ... Enable HTTP APIs server in `Features > HTTP APIs server`”
   • [claimed-docs] “we developed the local HTTP server to support external access to the database through authorization.”
   • [claimed-docs] “local HTTP server to access logseq's graph”

   proof ↗⚑ flag

   Reflect—none0/10

   Reflect's API docs describe OAuth2 with PKCE (no secret keys needed) for authenticating a client, but there is no mention of scopes, granular permissions, or the ability to issue least-privilege credentials limited to specific actions/data for an agent — the API is described as a general append/list interface, not a scoped credentialing system.

   • [claimed-docs] “For authentication, we support OAuth 2. You will need to create credentials here.”
   • [claimed-docs] “Our supports appending data to your notes, returning a list of the links you've bookmarked, and a few other things.”
   • [claimed-docs] “We do support OAuth's PKCE extension which means you can build a client to Reflect without shipping secret keys around.”

   proof ↗⚑ flag

8. ai-native userBuild against official SDKs

   weight 2 · round to Logseq
   Logseq~partialprobed6/10

   Logseq exposes a genuine plugin API/SDK (marketplace with hundreds of packages, load-unpacked plugin support) plus a local HTTP APIs server and a logseq:// URL protocol for third-party integration, giving AI-native builders real programmatic surfaces. However, the plugin API documentation is explicitly community-driven rather than official, and probes found no OpenAPI/machine-readable spec or llms.txt for these APIs. Missing for 10: official (not community-maintained) SDK documentation, an OpenAPI/schema spec, and evidence of first-party AI-agent integration guidance.

   • [claimed-docs] “Extends the functionality of logseq with third party tools ... third-party plugins from the Marketplace, can be searched, installed, updated…”
   • [claimed-docs] “local HTTP server to access logseq's graph ... Enable HTTP APIs server in `Features > HTTP APIs server`”
   • [claimed-docs] “This is a community-driven documentation for Logseq plugin APIs.”
   • [claimed-docs] “we developed the local HTTP server to support external access to the database through authorization.”
   • [claimed-docs] “Logseq provides a protocol `logseq://` to allow third party applications to invoke actions within Logseq.”
   • [probe] “PROBE runtime (recorded 2026-09-05): Logseq's official plugin marketplace registry (logseq/marketplace, one directory per package) counted 6…”
   • [probe] “PROBE llms.txt: HTTP 404 at https://plugins-doc.logseq.com/llms.txt”
   • [probe] “PROBE openapi: all candidate paths 404 (https://plugins-doc.logseq.com/openapi.json, https://plugins-doc.logseq.com/swagger.json, https://pl…”

   proof ↗⚑ flag

   Reflect~partialprobed3/10

   Reflect exposes a documented REST API with OAuth2/PKCE authentication that developers can build against, confirmed live via probe, but there is no dedicated SDK package, no OpenAPI/swagger spec, and no llms.txt for AI-native tooling — the API itself is limited (append-only, no note read due to E2E encryption). missing for 10: official SDK libraries, OpenAPI spec, llms.txt/AI-friendly docs, read access to note content.

   • [claimed-docs] “For authentication, we support OAuth 2. You will need to create credentials here.”
   • [claimed-docs] “Our supports appending data to your notes, returning a list of the links you've bookmarked, and a few other things.”
   • [claimed-docs] “We do support OAuth's PKCE extension which means you can build a client to Reflect without shipping secret keys around.”
   • [probe] “PROBE llms.txt: HTTP 404 at https://reflect.academy/llms.txt”
   • [probe] “PROBE openapi: all candidate paths 404 (https://reflect.academy/openapi.json, https://reflect.academy/swagger.json, https://reflect.academy/…”
   • [probe] “PROBE runtime (recorded 2026-09-05): keyless GET to the documented REST base https://reflect.app/api/graphs returned HTTP 401 {"error":{"typ…”

   proof ↗⚑ flag

9. ai-native userSubscribe to events via webhooks

   weight 2 · round drawn
   Logseq—none0/10

   Logseq offers a local HTTP API and a plugin system, but the evidence pack shows no webhook subscription mechanism (outbound event notifications) — only a local HTTP server for pull-based access and a custom URI protocol for inbound invocation. No documentation or community evidence describes subscribing to events via webhooks.

   ⚑ flag

   Reflect—none0/10

   Reflect's API docs describe OAuth2-based REST access for appending/reading data, but there is no mention of webhooks or event subscriptions anywhere in the docs or probes; the API is described as append-only and pull-based, not event-driven.

   • [claimed-docs] “For authentication, we support OAuth 2. You will need to create credentials here.”
   • [claimed-docs] “Our supports appending data to your notes, returning a list of the links you've bookmarked, and a few other things.”
   • [claimed-docs] “We do support OAuth's PKCE extension which means you can build a client to Reflect without shipping secret keys around.”
   • [probe] “PROBE runtime (recorded 2026-09-05): keyless GET to the documented REST base https://reflect.app/api/graphs returned HTTP 401 {"error":{"typ…”

   proof ↗⚑ flag

Agentic features

1. ai-native userGet AI-generated insights and suggestions from my data inside the product

   weight 2 · round to Reflect
   Logseq—none0/10

   Logseq's docs describe queries, plugins, journaling, sync, and export, but nowhere document a native AI-generated insights or suggestion feature. The only AI-adjacent mention is a community complaint that a user could no longer use Claude/codex to write/update pages, which points to third-party integration loss, not a first-party AI feature.

   • [community] “I've got so much already in logseq though, and I really like the model. Right now I'm thinking I'll just stay on version one as long as poss…”
   • [claimed-docs] “Simple queries by using `{{query }}`, the format is something like this: {{query Something you're looking for}}”
   • [claimed-docs] “Queries are for asking questions from your knowledge base and the outside world (in the future).”

   proof ↗⚑ flag

   Reflect✓fullclaimed7/10

   Reflect's AI palette lets users summarize meeting notes into action items/key takeaways, ask free-form questions, chat with filtered notes, and choose between GPT-4o and Claude 3.5 Sonnet — clear AI-generated insight/suggestion features built into the product. Missing for 10: independent/hands-on validation of insight quality and any proactive (vs. purely on-demand) suggestion generation.

   • [claimed-docs] “List action-items and key takeaways from meeting notes”
   • [claimed-docs] “You can also ask the AI palette anything you’d like using the open text field, and save it as a custom prompt to quickly call on again later…”
   • [claimed-docs] “To use it, first highlight some text, and then press `cmd j`... That’ll popup the AI palette.”
   • [claimed-docs] “You can also **ask the AI palette anything** you’d like using the open text field, and save it as a custom prompt to quickly call on again l…”
   • [claimed-docs] “Reflect's AI tools give you the option to now use either GPT-4o (the latest model from OpenAI) or Claude 3.5 Sonnet (the latest model from A…”
   • [claimed-docs] “you can also “chat” with those specific notes to ask questions and extract information.”
   • [claimed-docs] “After applying any desired filters to the search, you can also “chat” with those specific notes to ask questions and extract information.”

   proof ↗⚑ flag

2. ai-native userSet up automations that run autonomously in the background

   weight 2 · round drawn
   Logseq—none0/10

   Evidence shows Logseq has a plugin system, local HTTP API, and a custom URL protocol for third-party integration, but nothing describes autonomous, unattended background automations (e.g., schedulers, triggers, agent workflows) running without user action.

   • [claimed-docs] “Extends the functionality of logseq with third party tools ... third-party plugins from the Marketplace, can be searched, installed, updated…”
   • [claimed-docs] “local HTTP server to access logseq's graph ... Enable HTTP APIs server in `Features > HTTP APIs server`”
   • [claimed-docs] “we developed the local HTTP server to support external access to the database through authorization.”
   • [claimed-docs] “Logseq provides a protocol `logseq://` to allow third party applications to invoke actions within Logseq.”

   proof ↗⚑ flag

   Reflect—none0/10

   Reflect is a note-taking app with integrations, deep links, and an append-only API, but there is no evidence of any automation/workflow engine, scheduler, or trigger system that runs autonomously in the background without user action. Integrations (Trello, Asana, calendar sync) are passive data pulls, not user-configured autonomous automations.

   ⚑ flag

3. ai-native userDelegate tasks to a built-in AI assistant inside the product

   weight 3 · round to Reflect
   Logseq—none0/10

   No evidence in the docs or community pack of a built-in AI assistant inside Logseq; only third-party plugins, HTTP API, and protocol integrations exist, and one community comment references using external tools like Claude/Codex to edit pages, not a native assistant.

   • [claimed-docs] “Extends the functionality of logseq with third party tools ... third-party plugins from the Marketplace, can be searched, installed, updated…”
   • [claimed-docs] “local HTTP server to access logseq's graph ... Enable HTTP APIs server in `Features > HTTP APIs server`”
   • [community] “I've got so much already in logseq though, and I really like the model. Right now I'm thinking I'll just stay on version one as long as poss…”

   proof ↗⚑ flag

   Reflect~partialclaimed6/10

   Reflect ships a built-in 'AI palette' that can summarize notes, list action items, answer custom prompts, and let users 'chat' with their notes using GPT-4o or Claude 3.5 Sonnet — a real in-app AI assistant a user can delegate simple analysis/writing tasks to. However, this is limited to text generation/QA over notes rather than broader agentic task execution (no evidence of multi-step autonomous actions, tool use, or task automation beyond note content). Missing for 10: evidence of autonomous multi-step task delegation, agentic tool-use/actions beyond text Q&A, and independent hands-on verification of the AI assistant's reliability.

   • [claimed-docs] “List action-items and key takeaways from meeting notes”
   • [claimed-docs] “You can also ask the AI palette anything you’d like using the open text field, and save it as a custom prompt to quickly call on again later…”
   • [claimed-docs] “To use it, first highlight some text, and then press `cmd j`... That’ll popup the AI palette.”
   • [claimed-docs] “You can also **ask the AI palette anything** you’d like using the open text field, and save it as a custom prompt to quickly call on again l…”
   • [claimed-docs] “Reflect's AI tools give you the option to now use either GPT-4o (the latest model from OpenAI) or Claude 3.5 Sonnet (the latest model from A…”
   • [claimed-docs] “you can also “chat” with those specific notes to ask questions and extract information.”

   proof ↗⚑ flag

4. ai-native userOperate the product with natural-language commands

   weight 2 · round to Reflect
   Logseq—none0/10

   No evidence of a native natural-language command interface, chat-based control, or AI assistant in Logseq's docs; the product is described as a block/outline-based knowledge base with query syntax and plugin APIs, not NL commands. One community comment (logseq-comm-11) references losing the ability to use Claude/Codex to edit pages, implying such capability existed only via third-party plugin integration, not a first-party feature, and it's now broken in the new DB version.

   • [community] “I've got so much already in logseq though, and I really like the model. Right now I'm thinking I'll just stay on version one as long as poss…”
   • [claimed-docs] “Extends the functionality of logseq with third party tools ... third-party plugins from the Marketplace, can be searched, installed, updated…”
   • [claimed-docs] “Queries are for asking questions from your knowledge base and the outside world (in the future).”

   proof ↗⚑ flag

   Reflect~partialclaimed5/10

   Reflect's AI palette lets users type natural-language prompts to summarize, extract action items, or chat with notes (cmd-j palette, custom prompts), and search supports semantic natural-language queries. However, this NL support is limited to content Q&A/summarization rather than controlling the app's actions—actual app commands are exposed via structured deep-link URLs (reflect://...&command=...) rather than natural language, so full agentic NL control is not evidenced. Missing for 10: evidence of NL commands driving app actions (e.g., creating/organizing notes, triggering integrations) beyond querying/summarizing content, and any independent/hands-on confirmation of this NL interface's reliability.

   • [claimed-docs] “You can also ask the AI palette anything you’d like using the open text field, and save it as a custom prompt to quickly call on again later…”
   • [claimed-docs] “To use it, first highlight some text, and then press `cmd j`... That’ll popup the AI palette.”
   • [claimed-docs] “You can also **ask the AI palette anything** you’d like using the open text field, and save it as a custom prompt to quickly call on again l…”
   • [claimed-docs] “you can also “chat” with those specific notes to ask questions and extract information.”
   • [claimed-docs] “After applying any desired filters to the search, you can also “chat” with those specific notes to ask questions and extract information.”
   • [claimed-docs] “reflect://reflect?command=<command_name>&<parameter1>=<value1>&...”

   proof ↗⚑ flag

Api quality

1. ai-native userExplore an interactive API reference with runnable examples

   weight 2 · round drawn
   Logseq—none0/10

   There is a plugin API doc site and a local HTTP API, but no evidence of an interactive API reference with runnable examples; probes explicitly show no OpenAPI/swagger spec and no llms.txt found at the plugin docs site.

   • [claimed-docs] “This is a community-driven documentation for Logseq plugin APIs.”
   • [claimed-docs] “local HTTP server to access logseq's graph ... Enable HTTP APIs server in `Features > HTTP APIs server`”
   • [probe] “PROBE llms.txt: HTTP 404 at https://plugins-doc.logseq.com/llms.txt”
   • [probe] “PROBE openapi: all candidate paths 404 (https://plugins-doc.logseq.com/openapi.json, https://plugins-doc.logseq.com/swagger.json, https://pl…”

   proof ↗⚑ flag

   Reflect—none0/10

   Reflect's API docs (reflect.academy/api) describe OAuth2 endpoints and PKCE support in prose, but there is no evidence of an interactive API reference with runnable examples — probes confirm no OpenAPI/Swagger spec (404s) and no llms.txt, and nothing in the docs shows a live sandbox or try-it-now console.

   • [claimed-docs] “For authentication, we support OAuth 2. You will need to create credentials here.”
   • [claimed-docs] “Our supports appending data to your notes, returning a list of the links you've bookmarked, and a few other things.”
   • [claimed-docs] “We do support OAuth's PKCE extension which means you can build a client to Reflect without shipping secret keys around.”
   • [probe] “PROBE llms.txt: HTTP 404 at https://reflect.academy/llms.txt”
   • [probe] “PROBE openapi: all candidate paths 404 (https://reflect.academy/openapi.json, https://reflect.academy/swagger.json, https://reflect.academy/…”
   • [probe] “PROBE runtime (recorded 2026-09-05): keyless GET to the documented REST base https://reflect.app/api/graphs returned HTTP 401 {"error":{"typ…”

   proof ↗⚑ flag

2. ai-native userDownload a machine-readable API spec (OpenAPI or equivalent)

   weight 2 · round drawn
   Logseq—none0/10

   Logseq documents a local HTTP API server and a plugin API, but there is no evidence of a downloadable machine-readable spec (OpenAPI/Swagger) for either; explicit probes for openapi.json/swagger.json at the plugin docs site returned 404s.

   • [claimed-docs] “local HTTP server to access logseq's graph ... Enable HTTP APIs server in `Features > HTTP APIs server`”
   • [claimed-docs] “This is a community-driven documentation for Logseq plugin APIs.”
   • [probe] “PROBE llms.txt: HTTP 404 at https://plugins-doc.logseq.com/llms.txt”
   • [probe] “PROBE openapi: all candidate paths 404 (https://plugins-doc.logseq.com/openapi.json, https://plugins-doc.logseq.com/swagger.json, https://pl…”

   proof ↗⚑ flag

   Reflect—none0/10

   Reflect has an API with OAuth2/PKCE docs, but probes confirm no OpenAPI/Swagger spec or machine-readable schema is published (all candidate paths 404, llms.txt 404); documentation is prose-only.

   • [probe] “PROBE openapi: all candidate paths 404 (https://reflect.academy/openapi.json, https://reflect.academy/swagger.json, https://reflect.academy/…”
   • [probe] “PROBE llms.txt: HTTP 404 at https://reflect.academy/llms.txt”
   • [claimed-docs] “For authentication, we support OAuth 2. You will need to create credentials here.”
   • [claimed-docs] “We do support OAuth's PKCE extension which means you can build a client to Reflect without shipping secret keys around.”

   proof ↗⚑ flag

3. ai-native userTest against a sandbox environment without touching production data

   weight 1 · round drawn
   Logseq—none0/10

   The axis applies to this product kind (peer products hold positive or none verdicts on this story), so lack of evidence for an applicable capability is "none", never "na". (na/none harmonized at arena bring-up — see pipeline/scripts/na-harmonize.ts.)

   ⚑ flag

   Reflect—none0/10

   The axis applies to this product kind (peer products hold positive or none verdicts on this story), so lack of evidence for an applicable capability is "none", never "na". (na/none harmonized at arena bring-up — see pipeline/scripts/na-harmonize.ts.)

   ⚑ flag

4. ai-native userRely on versioned APIs with a documented deprecation policy

   weight 2 · round drawn
   Logseq—none0/10

   Logseq documents a local HTTP API and a plugin API, but there is no evidence of API versioning or a documented deprecation policy anywhere in the docs; probes for openapi/llms.txt specs return 404s, and community comments (e.g., DB-version migration breaking AI tool integrations) suggest breaking changes without clear versioning guarantees.

   • [claimed-docs] “local HTTP server to access logseq's graph ... Enable HTTP APIs server in `Features > HTTP APIs server`”
   • [claimed-docs] “we developed the local HTTP server to support external access to the database through authorization.”
   • [probe] “PROBE llms.txt: HTTP 404 at https://plugins-doc.logseq.com/llms.txt”
   • [probe] “PROBE openapi: all candidate paths 404 (https://plugins-doc.logseq.com/openapi.json, https://plugins-doc.logseq.com/swagger.json, https://pl…”
   • [community] “I've got so much already in logseq though, and I really like the model. Right now I'm thinking I'll just stay on version one as long as poss…”

   proof ↗⚑ flag

   Reflect—none0/10

   The API docs describe OAuth2/PKCE authentication and basic endpoints, but there is no evidence of any versioning scheme (e.g., v1/v2 paths), changelog, or documented deprecation policy anywhere in the docs or probes; OpenAPI/spec probes also 404. missing for 10: API version numbering, changelog/release notes, explicit deprecation policy or backward-compatibility guarantees.

   • [claimed-docs] “For authentication, we support OAuth 2. You will need to create credentials here.”
   • [claimed-docs] “Our supports appending data to your notes, returning a list of the links you've bookmarked, and a few other things.”
   • [claimed-docs] “We do support OAuth's PKCE extension which means you can build a client to Reflect without shipping secret keys around.”
   • [probe] “PROBE openapi: all candidate paths 404 (https://reflect.academy/openapi.json, https://reflect.academy/swagger.json, https://reflect.academy/…”
   • [probe] “PROBE runtime (recorded 2026-09-05): keyless GET to the documented REST base https://reflect.app/api/graphs returned HTTP 401 {"error":{"typ…”

   proof ↗⚑ flag

Agent ops

1. ai-native userMy coding agent or assistant can read AND write my notes headlessly — through a documented local API, vendor API, or MCP server — without me touching the GUI

   weight 3 · round to Reflect
   Logseq!disputedcontradicted5/10

   Logseq documents a local HTTP APIs server explicitly built 'to support external access to the database through authorization,' which would let an agent read/write notes headlessly, plus a plugin API and logseq:// protocol for third-party invocation. However, a hands-on community report directly contradicts this working capability, stating 'Not being able to use Claude or codex anymore to write or update pages is a real deal breaker for me' — indicating the API-based read/write access broke (likely in the newer DB version). Missing for 10: confirmation the HTTP API still supports write access post-DB-migration, an official MCP server, and independent corroboration that agent read/write works reliably today.

   • [claimed-docs] “local HTTP server to access logseq's graph ... Enable HTTP APIs server in `Features > HTTP APIs server`”
   • [claimed-docs] “local HTTP server to access logseq's graph”
   • [claimed-docs] “we developed the local HTTP server to support external access to the database through authorization.”
   • [claimed-docs] “Logseq provides a protocol `logseq://` to allow third party applications to invoke actions within Logseq.”
   • [community] “I've got so much already in logseq though, and I really like the model. Right now I'm thinking I'll just stay on version one as long as poss…”

   proof ↗⚑ flag

   Reflect~partialprobed4/10

   Reflect documents a real OAuth2/PKCE REST API (confirmed live via runtime probe) that supports appending data to notes and returning bookmarked links, giving agents a write path. However, the same probe notes the API is append-only by design because end-to-end encryption prevents Reflect's servers from reading note content — so headless READ of actual notes is not possible via the documented API, and no MCP server is documented. Missing for 10: a documented way to read/retrieve full note content headlessly, and an official MCP server integration.

   • [claimed-docs] “For authentication, we support OAuth 2. You will need to create credentials here.”
   • [claimed-docs] “Our supports appending data to your notes, returning a list of the links you've bookmarked, and a few other things.”
   • [claimed-docs] “We do support OAuth's PKCE extension which means you can build a client to Reflect without shipping secret keys around.”
   • [probe] “PROBE runtime (recorded 2026-09-05): keyless GET to the documented REST base https://reflect.app/api/graphs returned HTTP 401 {"error":{"typ…”
   • [claimed-docs] “The contents of your notes are end-to-end encrypted. No one else can read them (not even us).”

   proof ↗⚑ flag

Ai assist

1. ai-native userConnect external LLMs and AI tools to my notes through first-party connectors or an ecosystem of AI plugins

   weight 2 · round drawn
   Logseq~partialprobed5/10

   Logseq has no documented first-party AI/LLM connector, but its open plugin marketplace (617 registered packages), local HTTP API for external graph access, and logseq:// protocol for third-party app integration provide an ecosystem pathway for AI plugins/tools to connect to notes. Community evidence confirms some users did integrate LLMs like Claude/codex to write/update pages, but also that this broke in the newer DB version, showing fragility rather than a robust first-party AI integration story. Missing for 10: an official AI/LLM connector or documented AI plugin category, first-party docs mentioning LLM integration, and confirmation that AI plugin workflows are stable across versions.

   • [claimed-docs] “Extends the functionality of logseq with third party tools ... third-party plugins from the Marketplace, can be searched, installed, updated…”
   • [claimed-docs] “local HTTP server to access logseq's graph ... Enable HTTP APIs server in `Features > HTTP APIs server`”
   • [claimed-docs] “we developed the local HTTP server to support external access to the database through authorization.”
   • [claimed-docs] “Logseq provides a protocol `logseq://` to allow third party applications to invoke actions within Logseq.”
   • [probe] “PROBE runtime (recorded 2026-09-05): Logseq's official plugin marketplace registry (logseq/marketplace, one directory per package) counted 6…”
   • [community] “I've got so much already in logseq though, and I really like the model. Right now I'm thinking I'll just stay on version one as long as poss…”

   proof ↗⚑ flag

   Reflect~partialprobed5/10

   Reflect ships first-party AI features letting users toggle between GPT-4o and Claude 3.5 Sonnet for in-app AI palette actions (summarize, chat with notes), and exposes an OAuth2/PKCE API for building custom clients. However there's no evidence of a broader plugin ecosystem or first-party connectors letting external AI tools/agents read and act on notes — the API is documented as append-only and scoped narrowly (bookmarks, appending data), not an AI integration surface, and no third-party AI plugin marketplace is mentioned. Missing for 10: a documented AI-plugin ecosystem, evidence of external AI tools consuming notes via API, and richer API scope beyond append-only actions.

   • [claimed-docs] “Reflect's AI tools give you the option to now use either GPT-4o (the latest model from OpenAI) or Claude 3.5 Sonnet (the latest model from A…”
   • [claimed-docs] “To use it, first highlight some text, and then press `cmd j`... That’ll popup the AI palette.”
   • [claimed-docs] “You can also **ask the AI palette anything** you’d like using the open text field, and save it as a custom prompt to quickly call on again l…”
   • [claimed-docs] “For authentication, we support OAuth 2. You will need to create credentials here.”
   • [claimed-docs] “Our supports appending data to your notes, returning a list of the links you've bookmarked, and a few other things.”
   • [claimed-docs] “We do support OAuth's PKCE extension which means you can build a client to Reflect without shipping secret keys around.”
   • [probe] “PROBE runtime (recorded 2026-09-05): keyless GET to the documented REST base https://reflect.app/api/graphs returned HTTP 401 {"error":{"typ…”

   proof ↗⚑ flag

2. ai-native userA built-in AI assistant can answer questions, summarize, and draft grounded in my own notes — with clear controls over what data it sees

   weight 2 · round to Reflect
   Logseq—none0/10

   Logseq's own docs describe journals, queries, plugins, HTTP API, and sync, but there is no evidence of a built-in first-party AI assistant that answers/summarizes/drafts grounded in notes with data-visibility controls. A community comment even references losing the ability to use Claude/Codex on notes after a version change, implying any AI usage was via external plugins/tools rather than a native assistant.

   • [community] “I've got so much already in logseq though, and I really like the model. Right now I'm thinking I'll just stay on version one as long as poss…”
   • [claimed-docs] “Extends the functionality of logseq with third party tools ... third-party plugins from the Marketplace, can be searched, installed, updated…”
   • [claimed-docs] “local HTTP server to access logseq's graph ... Enable HTTP APIs server in `Features > HTTP APIs server`”

   proof ↗⚑ flag

   Reflect✓fullcommunity7/10

   Docs show a built-in AI palette (cmd-j) that can summarize meeting notes, list action items, answer custom prompts, and 'chat' with filtered/specific notes grounded in the user's own content, with model choice (GPT-4o/Claude 3.5). Data-scoping control is shown via search filters before chatting with a subset of notes, and E2E encryption is claimed for note content, though community evidence notes files/images are excluded from encryption. Missing for 10: independent hands-on review of AI answer quality/grounding, explicit UI controls for restricting AI's data access beyond search filters, and no clarification on how E2E encryption interacts with cloud-based AI processing.

   • [claimed-docs] “List action-items and key takeaways from meeting notes”
   • [claimed-docs] “you can also “chat” with those specific notes to ask questions and extract information.”
   • [claimed-docs] “You can also ask the AI palette anything you’d like using the open text field, and save it as a custom prompt to quickly call on again later…”
   • [claimed-docs] “After applying any desired filters to the search, you can also “chat” with those specific notes to ask questions and extract information.”
   • [claimed-docs] “To use it, first highlight some text, and then press `cmd j`... That’ll popup the AI palette.”
   • [claimed-docs] “You can also **ask the AI palette anything** you’d like using the open text field, and save it as a custom prompt to quickly call on again l…”
   • [claimed-docs] “Reflect's AI tools give you the option to now use either GPT-4o (the latest model from OpenAI) or Claude 3.5 Sonnet (the latest model from A…”
   • [claimed-docs] “Reflect’s Semantic search lets you find specific pieces of information without knowing the exact words.”
   • [claimed-docs] “The contents of your notes are end-to-end encrypted. No one else can read them (not even us).”
   • [community] “Question about Reflect's E2E encryption claims: 'We do not end-to-end encrypt things like files or images you upload' - user noted this excl…”

   proof ↗⚑ flag

1. ai-native userPerform bulk operations across many items at once

   weight 2 · round to Logseq
   Logseq~partialclaimed4/10

   Logseq exposes a local HTTP API and plugin system that could be used to script bulk edits across many blocks/pages, and its query language ({{query}}) can surface many items at once, but there is no documented native bulk-operation feature (e.g., mass rename, bulk tag, batch move) demonstrated in the evidence pack. missing for 10: explicit bulk-edit/bulk-update API or UI feature, evidence of an AI agent actually performing bulk operations via the HTTP API, independent hands-on confirmation of scale.

   • [claimed-docs] “Simple queries by using `{{query }}`, the format is something like this: {{query Something you're looking for}}”
   • [claimed-docs] “Simple queries by using `{{query }}`... You can type `/query` to create a simple query.”
   • [claimed-docs] “Queries are for asking questions from your knowledge base and the outside world (in the future).”
   • [claimed-docs] “local HTTP server to access logseq's graph ... Enable HTTP APIs server in `Features > HTTP APIs server`”
   • [claimed-docs] “we developed the local HTTP server to support external access to the database through authorization.”
   • [claimed-docs] “This is a community-driven documentation for Logseq plugin APIs.”

   proof ↗⚑ flag

   Reflect—none0/10

   Reflect's API supports appending data to individual notes and returning bookmark lists, but there is no evidence of any bulk/batch operation capability (e.g., batch update, multi-note transformation, or scripted bulk edits across many items at once); the API is explicitly append-only per note due to E2E encryption.

   • [claimed-docs] “Our supports appending data to your notes, returning a list of the links you've bookmarked, and a few other things.”
   • [claimed-docs] “We do support OAuth's PKCE extension which means you can build a client to Reflect without shipping secret keys around.”
   • [probe] “PROBE runtime (recorded 2026-09-05): keyless GET to the documented REST base https://reflect.app/api/graphs returned HTTP 401 {"error":{"typ…”

   proof ↗⚑ flag

2. ai-native userDefine rules that trigger actions automatically on events

   weight 3 · round drawn
   Logseq—none0/10

   Logseq's docs describe journaling, queries, plugins, HTTP APIs, and protocol handlers, but nowhere document a rules/trigger system where user-defined events automatically fire actions (e.g., 'when block tagged X, do Y'). The plugin API and marketplace could theoretically enable such automation, but no evidence shows a first-party or documented feature for event-triggered rules.

   • [claimed-docs] “Extends the functionality of logseq with third party tools ... third-party plugins from the Marketplace, can be searched, installed, updated…”
   • [claimed-docs] “This is a community-driven documentation for Logseq plugin APIs.”
   • [claimed-docs] “Logseq provides a protocol `logseq://` to allow third party applications to invoke actions within Logseq.”

   proof ↗⚑ flag

   Reflect—none0/10

   Reflect's docs describe integrations (Trello/Asana/calendar sync), deep links, and an API, but nowhere is there a rules engine or event-trigger/action automation system (e.g., 'when X happens, do Y'). No evidence supports this capability.

   ⚑ flag

Capture

1. researcherClip web pages, articles, and highlights from my browser straight into my notes with an official clipper or capture flow

   weight 2 · round to Reflect
   Logseq~partialclaimed4/10

   Logseq documents a 'Quick capture' feature via the logseq:// protocol that can be added as a browser bookmarklet or invoked from command line, letting users capture short text and links into notes (logseq-docs-11, logseq-docs-17, logseq-docs-27, logseq-docs-22). However, this is a lightweight bookmarklet, not a dedicated official browser-extension web clipper capable of capturing full articles, page content, or highlights as many competing tools offer. Missing for 10: an official browser extension/clipper product, support for capturing full article content or highlighted text with metadata, and any independent/hands-on evidence of researchers using this capture flow for web clipping.

   • [claimed-docs] “logseq://x-callback-url/quickCapture? ... Quick capture for short text and links. Can be added as a bookmark link to modern browser”
   • [claimed-docs] “logseq://x-callback-url/quickCapture? - Quick capture for short text and links.”
   • [claimed-docs] “Logseq provides a protocol `logseq://` to allow third party applications to invoke actions within Logseq.”
   • [claimed-docs] “Quick capture for short text and links. Can be added as a bookmark link to modern browser or opened from command line shell”

   proof ↗⚑ flag

   Reflect✓fullclaimed9/10

   Reflect ships an official Chrome and Safari browser extension for bookmarking websites, capturing snippets, and highlights, which are automatically added to the daily note — directly matching the researcher's clipper/capture workflow. Documentation is detailed and first-party; missing for 10: independent hands-on review specifically validating the clipper extension's reliability/quality.

   • [claimed-docs] “Reflect lets you quickly bookmark websites and capture snippets from across the web using our Chrome extension”
   • [claimed-docs] “The link and your highlights will automatically be added to your daily note.”
   • [claimed-docs] “Reflect lets you quickly bookmark websites and capture snippets from across the web using our Chrome extension and Safari extension.”

   proof ↗⚑ flag

Daily notes

1. power-userI get an automatic daily note for each day, pre-filled from a template, as the default landing place for journaling and capture

   weight 3 · round drawn
   Logseq✓fullclaimed8/10

   Docs confirm Logseq opens to the Journals page every time you launch it ([logseq-docs-24]), that a new dated journal page is auto-created daily at midnight ([logseq-docs-2]), and that this page serves as the equivalent of a daily-notes capture landing page ([logseq-docs-1]). Missing for 10: explicit documentation of a customizable 'journal template' pre-fill mechanism and independent/hands-on confirmation that templates apply automatically to new daily notes.

   • [claimed-docs] “Every time you open Logseq, you start on the Journals page.”
   • [claimed-docs] “Every day at midnight a new Journals page is created with that day's date, making it easy to see when you've added what information.”
   • [claimed-docs] “For people familiar with Roam Research: the Journals page in Logseq is the same as the Daily Notes Page in Roam.”

   proof ↗⚑ flag

   Reflect✓fullclaimed8/10

   Reflect's docs explicitly confirm automatic daily notes created every day as the journaling landing place, and multiple features (voice transcription, bookmarking, calendar sync, integrations) are documented as auto-filling into that daily note, showing it's the default capture surface. Missing for 10: explicit confirmation of a customizable daily-note *template* pre-filling structure/content, and independent hands-on verification of the daily-note experience beyond docs.

   • [claimed-docs] “Tools like Reflect will create a new _daily note_ for you everyday for you to journal in.”
   • [claimed-docs] “Tools like Reflect will create a new daily note for you everyday for you to journal in.”
   • [claimed-docs] “Speak your note, then click stop recording... their transcription will be added to your daily note.”
   • [claimed-docs] “The link and your highlights will automatically be added to your daily note.”
   • [claimed-docs] “We support connecting your Google, iCloud and Office 365 account(s) to Reflect. This lets you: Pull in your calendar events, so you can easi…”
   • [claimed-docs] “Add Trello cards or Asana tasks to your Daily Note”
   • [claimed-docs] “Add Trello cards or Asana tasks to your Daily Note Add Google Docs and Slack notes to Reflect Save emails and Tweets to your notes Sync your…”

   proof ↗⚑ flag

Editor

1. power-userDefine reusable templates (with variables like date and title) and apply them when creating recurring kinds of notes

   weight 2 · round drawn
   Logseq—none0/10

   The evidence pack contains no documentation or community mentions of Logseq's template feature (e.g. `template::` property, `/template` command, or variable substitution like date/title) — only journals, page/block references, queries, plugins, sync, and publishing are covered. Since template creation is a plausible and well-known capability for a note-taking tool, its complete absence from the evidence means this axis cannot be credited as delivered.

   ⚑ flag

   Reflect—none0/10

   The evidence pack shows daily notes, backlinks, tags, filters, imports, and AI palette features, but nowhere mentions reusable note templates with variable substitution (e.g., date/title placeholders) for creating recurring note types. Missing for 10: any documentation of a templates feature, variable/placeholder syntax, or ability to save/apply templates to new notes.

   ⚑ flag

2. power-userWrite and format notes with headings, callouts, tables, code blocks, and embedded media in a fast keyboard-driven editor

   weight 2 · round to Logseq
   Logseq!disputedcontradicted4/10

   The evidence pack lacks any first-party documentation specifically describing headings, callouts, tables, code blocks, or media-embedding support, and instead shows only adjacent editor features (page/block references, journals) that imply block-based markdown editing. Concrete hands-on reports contradict the 'fast, keyboard-driven' claim: pasting multi-paragraph text broke formatting into unwanted lists and even crashed on longer pastes, and users report severe UI lag once graphs grow, undermining the 'fast' editing experience for power users. Missing for 10: explicit docs on headings/callouts/tables/code-block syntax, embedded media handling, and independent confirmation that these specific formatting features perform well at scale.

   • [claimed-docs] “Type `[[]]` to create or reference any other page. You can also make a new page in the search bar. - Type `(())` to reference a block on any…”
   • [claimed-docs] “Logseq: A privacy-first, open-source knowledge base”
   • [community] “I used Logseq for a while but the big problem I had with it is how strongly it is focused on bullet points. Posting long form text is cumber…”
   • [community] “Admittedly poor performance has been my primary gripe in the 7mos I've been using LogSeq on a near-daily basis... I'd gladly pay for a subsc…”
   • [community] “Indexing and general responsiveness slows down massively once you have a few million words stored, and you have to break up your data into m…”

   proof ↗⚑ flag

   Reflect—none0/10

   The evidence pack covers linking/backlinks, daily notes, import/export in Markdown, AI features, and encryption, but contains no documentation or community evidence about core rich-text editing capabilities such as headings, callouts, tables, code blocks, or embedded media, nor about keyboard-driven editing speed. Missing for 10: any docs mentioning headings/callout blocks/tables/code blocks/embedded media support, and any evidence of keyboard-shortcut-driven formatting workflow.

   ⚑ flag

Migration

1. switcherExport my entire knowledge base at any time in open formats, so leaving is as easy as arriving

   weight 2 · round to Reflect
   Logseq!disputedcontradicted5/10

   Logseq documents a genuine multi-format Export feature (Markdown, HTML, OPML, JSON, EDN, PNG) and historically stores data as local plain Markdown files, supporting easy exit at any time [logseq-docs-8]. However, hands-on community reports contradict the 'open, portable' framing: exported Markdown adds non-standard formatting that breaks compatibility with other Markdown apps like Obsidian [logseq-comm-19], and the newer Logseq DB version reportedly moves away from storing everything as plain markdown files, alarming users who relied on that for portability [logseq-comm-14, logseq-comm-9]. Missing for 10: independent verification that exports are lossless/round-trippable, clarity on whether DB-version graphs can still export to plain markdown, and confirmation the export covers all data (assets, whiteboards, plugins config).

   • [claimed-docs] “The Logseq Export feature allows users to export their data in various formats, including text, HTML, OPML, PNG, EDN, JSON, and standard Mar…”
   • [community] “The MarkDown files it generates aren't compatible with other similar programs (Obsidian for instance). It adds 2 spaces which makes things i…”
   • [community] “Logseq remained a buggy mess, is now on an unmaintained (thus insecure) version of electron. And now after several years of complete stagnat…”
   • [community] “Yeah I felt the same about the decision when they decided to move to database structure. If it had this structure from the start it could've…”

   proof ↗⚑ flag

   Reflect✓fullcommunity8/10

   Reflect explicitly documents export into open formats (HTML, Markdown) at any time, and the founder corroborates this as an anti-lock-in feature in community discussion. missing for 10: independent hands-on verification of a full/bulk knowledge-base export (vs per-note export) and confirmation that all note types/attachments are included.

   • [claimed-docs] “you can always export it into an open format (like HTML or Markdown).”
   • [community] “Reflect founder on data lock-in: 'you can always export your notes into markdown or HTML so you shouldn't feel too locked in.' Also noted th…”

   proof ↗⚑ flag

2. switcherImport my existing notes from Notion, Evernote, Roam, or plain Markdown with structure, links, and attachments preserved

   weight 3 · round to Reflect
   Logseq—none0/10

   The evidence pack documents Logseq's export formats and journal/page structure comparisons to Roam, but contains no mention of an import feature for Notion, Evernote, Roam exports, or Markdown migration with links/attachments preserved. Community comments mention switching from Roam via GitHub sync, not a structured import tool. Since import tooling is a plausible feature for a PKM app, absence of evidence means 'none' rather than 'na'.

   • [claimed-docs] “The Logseq Export feature allows users to export their data in various formats, including text, HTML, OPML, PNG, EDN, JSON, and standard Mar…”
   • [claimed-docs] “For people familiar with Roam Research: the Journals page in Logseq is the same as the Daily Notes Page in Roam.”
   • [community] “I switched over to LogSeq from Roam Research, and I'm using Github for syncing and backup.”

   proof ↗⚑ flag

   Reflect~partialclaimed6/10

   Reflect's docs confirm native import support for Evernote (ENEX), Roam (JSON), and Markdown files, three of the four sources named in the story, plus Apple Notes and Workflowy. However, Notion import is not mentioned anywhere in the import/export documentation, and there is no explicit confirmation that internal links or attachments are preserved through the import process for any of these formats. Missing for 10: Notion import support, explicit evidence that links and attachments (not just note text) survive import.

   • [claimed-docs] “We currently support imports from: Apple Notes (HTML) Evernote (ENEX) Markdown files Roam (JSON) Workflowy (OPML)”
   • [claimed-docs] “We currently support imports from: * Apple Notes (HTML) * Evernote (ENEX) * Markdown files * Roam (JSON) * Workflowy (OPML)”
   • [claimed-docs] “First you’ll have to export your existing Apple Notes using the provided script below, then import them into Reflect normally.”
   • [claimed-docs] “you can always export it into an open format (like HTML or Markdown).”

   proof ↗⚑ flag

Graph

1. researcherExplore my knowledge base as an interactive graph of notes and connections, filtered to what I care about

   weight 1 · round to Logseq
   Logseq~partialcommunity5/10

   Logseq's documented linking primitives — page/block references ([[ ]], (( ))) and simple queries ({{query}}) — provide the backbone for connecting and filtering notes, and the Whiteboard feature offers a spatial canvas with connectors for arranging knowledge visually, but no evidence explicitly documents a dedicated interactive 'graph view' with node-link visualization or filtering controls. Community reports also note serious performance degradation ('indexing and general responsiveness slows down massively once you have a few million words') which would hamper exploring a large graph. missing for 10: explicit documentation of the graph-view visualization UI and its filtering options, and independent confirmation that graph exploration performs well at scale.

   • [claimed-docs] “Type `[[]]` to create or reference any other page. You can also make a new page in the search bar. - Type `(())` to reference a block on any…”
   • [claimed-docs] “Simple queries by using `{{query }}`, the format is something like this: {{query Something you're looking for}}”
   • [claimed-docs] “Simple queries by using `{{query }}`... You can type `/query` to create a simple query.”
   • [claimed-docs] “Queries are for asking questions from your knowledge base and the outside world (in the future).”
   • [claimed-docs] “Put thoughts from your knowledge base as well as new ones next to each other on a spatial canvas together with shapes, drawings, website emb…”
   • [community] “Indexing and general responsiveness slows down massively once you have a few million words stored, and you have to break up your data into m…”

   proof ↗⚑ flag

   Reflect—none0/10

   Reflect's docs describe backlinks, [[ ]] linking, and filtering notes by tags/date/backlinks, but nowhere is an interactive visual graph view of the knowledge base mentioned or shown — the evidence only covers list-style backlinks and search filters, not a graph exploration UI.

   • [claimed-docs] “Networked note-taking tools, like Reflect, support creating links between notes... backlinks”
   • [claimed-docs] “Backlinks are just a list of incoming links from other notes.”
   • [claimed-docs] “You can filter notes by: Date created or updated, Daily Notes, Where they are backlinked to/from, #tags, Pinned notes”
   • [claimed-docs] “You can filter notes by: Date created or updated Daily Notes Where they are backlinked to/from #tags Pinned notes”
   • [claimed-docs] “You can filter notes by: * Date created or updated * Daily Notes * Where they are backlinked to/from * #tags * Pinned notes”

   proof ↗⚑ flag

Links

1. researcherLink notes with [[wiki-style]] links and see every backlink and unlinked mention of a note in context

   weight 3 · round drawn
   Logseq~partialcommunity7/10

   Docs confirm core wiki-linking mechanics ([[ ]] page references, (( )) block references, journals as the default landing page) and queries for surfacing related content, which underpins backlinking functionality core to Logseq's outliner design. However, the evidence pack never explicitly documents a dedicated 'Linked References' or 'unlinked references' UI panel showing backlinks-in-context, which is a well-known Logseq feature but unevidenced here; community notes instead focus on performance/sync issues rather than confirming backlink display quality. Missing for 10: explicit docs/evidence on the Linked References and Unlinked References panels, independent hands-on confirmation of backlink context display working well at scale.

   • [claimed-docs] “Type `[[]]` to create or reference any other page. You can also make a new page in the search bar. - Type `(())` to reference a block on any…”
   • [claimed-docs] “Simple queries by using `{{query }}`, the format is something like this: {{query Something you're looking for}}”
   • [claimed-docs] “Simple queries by using `{{query }}`... You can type `/query` to create a simple query.”
   • [claimed-docs] “Queries are for asking questions from your knowledge base and the outside world (in the future).”
   • [claimed-docs] “Every time you open Logseq, you start on the Journals page.”
   • [community] “Indexing and general responsiveness slows down massively once you have a few million words stored, and you have to break up your data into m…”

   proof ↗⚑ flag

   Reflect~partialclaimed7/10

   Docs confirm [[ ]] wiki-link shortcut, backlinks as incoming links, and filtering notes by where they're backlinked to/from, covering the core linking-graph story. However, evidence never explicitly mentions detection/surfacing of 'unlinked mentions' in context, which is a specific part of the story. Missing for 10: explicit unlinked-mention detection/surfacing, independent hands-on demonstration of backlink UI in context.

   • [claimed-docs] “Networked note-taking tools, like Reflect, support creating links between notes... backlinks”
   • [claimed-docs] “Generally note-taking tools, including Reflect, use the `[[` shortcut to link some text.”
   • [claimed-docs] “Backlinks are just a list of incoming links from other notes.”
   • [claimed-docs] “You can filter notes by: Date created or updated, Daily Notes, Where they are backlinked to/from, #tags, Pinned notes”
   • [claimed-docs] “You can filter notes by: Date created or updated Daily Notes Where they are backlinked to/from #tags Pinned notes”
   • [claimed-docs] “You can filter notes by: * Date created or updated * Daily Notes * Where they are backlinked to/from * #tags * Pinned notes”

   proof ↗⚑ flag

2. power-userReference or transclude an individual block or section of one note inside another and have it stay in sync

   weight 2 · round to Logseq
   Logseq✓fullclaimed8/10

   Logseq's block reference syntax `(())` embeds/transcludes an individual block from any page into another, and page references `[[]]` do the same at page level; since Logseq's block-based architecture keeps a single canonical block that references simply render/link to, edits propagate automatically, keeping transclusions in sync. Docs explicitly describe both block and page referencing as core features. Missing for 10: no explicit doc quote confirming live-sync behavior on edit (implied by architecture, not directly demonstrated) and no independent hands-on confirmation of transclusion sync specifically.

   • [claimed-docs] “Type `[[]]` to create or reference any other page. You can also make a new page in the search bar. - Type `(())` to reference a block on any…”
   • [claimed-docs] “Blocks that have a page as a parent i.e. usually top-level blocks can be converted to a page and back to a block easily. To convert this blo…”

   proof ↗⚑ flag

   Reflect—none0/10

   Evidence shows Reflect supports note-to-note linking and backlinks ([[ shortcuts, backlinks list) but nowhere documents block- or section-level transclusion/embedding that stays synced across notes — only whole-note linking and filtering by backlink is described.

   • [claimed-docs] “Networked note-taking tools, like Reflect, support creating links between notes... backlinks”
   • [claimed-docs] “Generally note-taking tools, including Reflect, use the `[[` shortcut to link some text.”
   • [claimed-docs] “Backlinks are just a list of incoming links from other notes.”

   proof ↗⚑ flag

Ownership

1. power-userMy notes live in local files or a local-first store that I own and can read without the app, and everything keeps working fully offline

   weight 3 · round to Logseq
   Logseq~partialcommunity6/10

   Logseq's classic (file) mode stores notes as plain Markdown on local disk, exports to multiple open formats, and only publishes pages explicitly (docs-8, docs-7, docs-23), supporting the 'own your files, work offline' story; community users confirm they read/back up the raw Markdown files themselves (comm-1, comm-19, comm-7). However, the newer DB-based version raises real doubts: a user explicitly states 'I can no longer keep all my data as markdown files' with the new format (comm-14), and others cite years of stagnation, sync bugs, and performance/data-loss concerns (comm-3, comm-5, comm-6, comm-8) that undercut confidence in durable, app-independent local ownership going forward. missing for 10: clear first-party confirmation that the DB version still exposes plain-text/local files, independent verification of offline reliability at scale, and resolution of the data-loss/performance complaints.

   • [claimed-docs] “The Logseq Export feature allows users to export their data in various formats, including text, HTML, OPML, PNG, EDN, JSON, and standard Mar…”
   • [claimed-docs] “By default no pages are public. Individual pages can be marked as public with the property `publish:: true`.”
   • [claimed-docs] “Logseq: A privacy-first, open-source knowledge base”
   • [community] “I switched over to LogSeq from Roam Research, and I'm using Github for syncing and backup.”
   • [community] “The MarkDown files it generates aren't compatible with other similar programs (Obsidian for instance). It adds 2 spaces which makes things i…”
   • [community] “I use Logseq personally and for work almost daily since a year ago or so. It's pretty nice and the 'least bad' PKM platform I'm aware of (no…”
   • [community] “Logseq remained a buggy mess, is now on an unmaintained (thus insecure) version of electron. And now after several years of complete stagnat…”
   • [community] “I had two main problems with LogSeq: 1. Not designed from the ground-up to prevent data loss. Has had data-loss issues in the past.”
   • [community] “Admittedly poor performance has been my primary gripe in the 7mos I've been using LogSeq on a near-daily basis... I'd gladly pay for a subsc…”

   proof ↗⚑ flag

   Reflect—none0/10

   Reflect is a cloud-synced, end-to-end-encrypted note service (docs-16/35) with export-to-Markdown/HTML available (docs-44, comm-7), but there is no evidence notes are stored as local files or in a local-first store the user can read without the app, nor evidence of full offline functionality — the API and encryption model imply server-mediated sync rather than local-first ownership.

   • [claimed-docs] “End-to-end encryption keeps your notes secure.”
   • [claimed-docs] “The contents of your notes are end-to-end encrypted. No one else can read them (not even us).”
   • [claimed-docs] “you can always export it into an open format (like HTML or Markdown).”
   • [community] “Reflect founder on data lock-in: 'you can always export your notes into markdown or HTML so you shouldn't feel too locked in.' Also noted th…”
   • [community] “Question about Reflect's E2E encryption claims: 'We do not end-to-end encrypt things like files or images you upload' - user noted this excl…”

   proof ↗⚑ flag

2. researcherMy notes are stored or exportable in open, documented formats (Markdown, JSON) so my archive outlives the vendor

   weight 2 · round to Reflect
   Logseq!disputedcontradicted5/10

   Logseq's docs confirm export to open, documented formats (Markdown, JSON, EDN, OPML) and its journal-based local storage model, supporting archive portability (logseq-docs-8). However, hands-on community reports contradict full open-format durability: users note the generated Markdown isn't standard-compliant/interoperable with other tools due to extra indentation (logseq-comm-19), and the newer DB-based version reportedly abandons plain Markdown as the storage format entirely, forcing users into a proprietary DB structure (logseq-comm-14). Missing for 10: confirmation that the current/default file storage (not just export) remains plain Markdown across versions, and independent verification that exported files are lossless/interoperable.

   • [claimed-docs] “The Logseq Export feature allows users to export their data in various formats, including text, HTML, OPML, PNG, EDN, JSON, and standard Mar…”
   • [community] “The MarkDown files it generates aren't compatible with other similar programs (Obsidian for instance). It adds 2 spaces which makes things i…”
   • [community] “Logseq remained a buggy mess, is now on an unmaintained (thus insecure) version of electron. And now after several years of complete stagnat…”

   proof ↗⚑ flag

   Reflect✓fullcommunity8/10

   Reflect explicitly documents export to open formats (Markdown, HTML) and import from Markdown/JSON-based formats, and the founder publicly confirms notes can always be exported to Markdown/HTML to avoid lock-in. Missing for 10: no independent hands-on verification of export fidelity/completeness, and export format is somewhat limited (HTML/Markdown only, not full JSON archive of all metadata).

   • [claimed-docs] “We currently support imports from: Apple Notes (HTML) Evernote (ENEX) Markdown files Roam (JSON) Workflowy (OPML)”
   • [claimed-docs] “We currently support imports from: * Apple Notes (HTML) * Evernote (ENEX) * Markdown files * Roam (JSON) * Workflowy (OPML)”
   • [claimed-docs] “you can always export it into an open format (like HTML or Markdown).”
   • [community] “Reflect founder on data lock-in: 'you can always export your notes into markdown or HTML so you shouldn't feel too locked in.' Also noted th…”

   proof ↗⚑ flag

Self host

1. developerRun the whole system without the vendor's cloud — local-only mode or a self-hosted sync backend I operate myself

   weight 2 · round to Logseq
   Logseq✓fullcommunity7/10

   Logseq stores data as local Markdown/EDN files by default with no cloud requirement, and is documented as 'privacy-first' — users confirm they run it fully local or pair it with self-managed sync like GitHub instead of vendor cloud. The official Logseq Sync exists only as an optional paid vendor-hosted add-on, not a self-hosted backend, and one user explicitly notes there's no built-in way to self-host it as a server despite an open feature request. missing for 10: an officially supported self-hosted sync backend (only vendor-cloud Sync or DIY git workflows exist), and independent verification that local mode has no offline degradation.

   • [claimed-docs] “Logseq: A privacy-first, open-source knowledge base”
   • [claimed-docs] “Syncs up to 10 graphs between Logseq Desktop, Android, and iOS.”
   • [claimed-docs] “Yes, your graph is end-to-end encrypted.”
   • [community] “I switched over to LogSeq from Roam Research, and I'm using Github for syncing and backup.”
   • [community] “I use Logseq personally and for work almost daily since a year ago or so. It's pretty nice and the 'least bad' PKM platform I'm aware of (no…”
   • [community] “I know Logseq is intended as an offline-first solution, but I really wish I could self-host it as a website and have all notes saved on my s…”

   proof ↗⚑ flag

   Reflect—none0/10

   Reflect is a cloud-hosted, closed-source SaaS with E2E encryption on the vendor's servers — there is no evidence of a local-only mode or self-hostable sync backend; only export to Markdown/HTML for backup is documented, not operating your own sync server.

   • [claimed-docs] “you can always export it into an open format (like HTML or Markdown).”
   • [community] “Reflect founder on data lock-in: 'you can always export your notes into markdown or HTML so you shouldn't feel too locked in.' Also noted th…”
   • [claimed-docs] “The contents of your notes are end-to-end encrypted. No one else can read them (not even us).”

   proof ↗⚑ flag

Mobile

1. power-userFull-featured iOS and Android apps give me my whole knowledge base — capture, search, and linking — not a stripped-down companion

   weight 2 · round drawn
   Logseq—none0/10

   Evidence confirms Logseq has mobile apps (iOS/Android sync up to 10 graphs, end-to-end encrypted) but contains no evidence about the actual feature set of the mobile apps themselves — no mention of whether search, linking, queries, or full editing capabilities are present on mobile versus a stripped-down capture-only experience. Community commentary focuses on desktop performance/sync bugs, not mobile app completeness.

   • [claimed-docs] “Syncs up to 10 graphs between Logseq Desktop, Android, and iOS.”
   • [claimed-docs] “Yes, your graph is end-to-end encrypted.”

   proof ↗⚑ flag

   Reflect—none0/10

   The evidence pack covers imports, API, AI features, publishing, security, and integrations extensively, but contains no mention of iOS or Android apps, their feature set, or mobile-specific capture/search/linking capabilities. Without any documentation or community evidence describing dedicated mobile apps (full-featured or otherwise), there is nothing to substantiate this story.

   ⚑ flag

1. ai-native userDo everything through the API that I can do in the UI

   weight 2 · round to Reflect
   Logseq!disputedcontradicted4/10

   Logseq docs describe a local HTTP APIs server for external access to the graph, a community-documented plugin API, and a logseq:// protocol for third-party invocation, suggesting programmatic access beyond the UI (logseq-docs-6, logseq-docs-21, logseq-docs-12, logseq-docs-22). However, there's no documentation showing this API surface covers 'everything' the UI can do (queries, whiteboards, publishing, sync are UI-first features with no documented API equivalents), and a hands-on user reports that after the move to the new DB version, 'Not being able to use Claude or codex anymore to write or update pages is a real deal breaker' — a concrete failure of AI-driven read/write parity (logseq-comm-11). missing for 10: full API coverage of queries/whiteboard/publish/sync, evidence of AI tools reliably reading+writing via API in the current version, independent confirmation the HTTP API matches UI capability breadth.

   • [claimed-docs] “local HTTP server to access logseq's graph ... Enable HTTP APIs server in `Features > HTTP APIs server`”
   • [claimed-docs] “we developed the local HTTP server to support external access to the database through authorization.”
   • [claimed-docs] “This is a community-driven documentation for Logseq plugin APIs.”
   • [claimed-docs] “Logseq provides a protocol `logseq://` to allow third party applications to invoke actions within Logseq.”
   • [community] “I've got so much already in logseq though, and I really like the model. Right now I'm thinking I'll just stay on version one as long as poss…”

   proof ↗⚑ flag

   Reflect~partialprobed3/10

   Reflect does expose an OAuth2/PKCE REST API confirmed live via runtime probe, but by its own docs it only supports appending data to notes and returning bookmarked links plus 'a few other things' — a small subset of UI capabilities (semantic search, AI chat, publishing, backlinks, calendar/integrations, deep links) with no evidence these are API-accessible. The API is also explicitly append-only for note content since E2E encryption prevents server-side reads, meaning even reading/editing existing notes via API is not possible as it is in the UI. Missing for 10: API endpoints for search/AI features, note reading/editing, publishing, and integrations parity with the UI.

   • [claimed-docs] “For authentication, we support OAuth 2. You will need to create credentials here.”
   • [claimed-docs] “Our supports appending data to your notes, returning a list of the links you've bookmarked, and a few other things.”
   • [claimed-docs] “We do support OAuth's PKCE extension which means you can build a client to Reflect without shipping secret keys around.”
   • [probe] “PROBE runtime (recorded 2026-09-05): keyless GET to the documented REST base https://reflect.app/api/graphs returned HTTP 401 {"error":{"typ…”
   • [claimed-docs] “Reflect’s Semantic search lets you find specific pieces of information without knowing the exact words.”
   • [claimed-docs] “Reflect offers one-click publishing of your notes. Simply hit the publish button and your note will be publicly available behind a long hard…”

   proof ↗⚑ flag

2. ai-native userExport all of my data in open formats and leave

   weight 3 · round to Reflect
   Logseq~partialcommunity6/10

   Logseq's docs describe a native local-markdown storage model plus an explicit Export feature covering text, HTML, OPML, PNG, EDN, JSON and standard Markdown, which supports data portability and offline 'leave anytime' use (logseq-docs-8). However, community reports note markdown output isn't cleanly interoperable with other markdown apps due to formatting quirks (logseq-comm-19), and the newer DB-based version is criticized for moving away from plain markdown files as the source of truth, undermining the 'leave in open formats' guarantee for that version (logseq-comm-14). Missing for 10: independent verification that exported formats round-trip cleanly into other tools, and clarity on export fidelity in the new DB architecture.

   • [claimed-docs] “The Logseq Export feature allows users to export their data in various formats, including text, HTML, OPML, PNG, EDN, JSON, and standard Mar…”
   • [community] “The MarkDown files it generates aren't compatible with other similar programs (Obsidian for instance). It adds 2 spaces which makes things i…”
   • [community] “Logseq remained a buggy mess, is now on an unmaintained (thus insecure) version of electron. And now after several years of complete stagnat…”

   proof ↗⚑ flag

   Reflect✓fullcommunity8/10

   Reflect explicitly documents export to open formats (HTML, Markdown) and the founder confirms this is a deliberate anti-lock-in policy, directly matching the story. missing for 10: no independent hands-on verification of a full data export/migration, and export is confined to notes (E2E-encrypted files/images may not be fully portable per reflect-comm-8).

   • [claimed-docs] “you can always export it into an open format (like HTML or Markdown).”
   • [claimed-docs] “We currently support imports from: Apple Notes (HTML) Evernote (ENEX) Markdown files Roam (JSON) Workflowy (OPML)”
   • [community] “Reflect founder on data lock-in: 'you can always export your notes into markdown or HTML so you shouldn't feel too locked in.' Also noted th…”
   • [community] “Question about Reflect's E2E encryption claims: 'We do not end-to-end encrypt things like files or images you upload' - user noted this excl…”

   proof ↗⚑ flag

3. ai-native userRead the product's source under an open license

   weight 2 · round to Logseq
   Logseq✓fullcommunity8/10

   Logseq is explicitly described as "open-source" (logseq-docs-23), and its docs, plugin ecosystem, and marketplace are hosted publicly on GitHub, implying the source is inspectable and openly licensed; community comments corroborate "local data and source availability are important factors" (logseq-comm-7). missing for 10: explicit citation of the actual license name/terms (e.g., AGPL) and any independent audit or discussion of license compliance.

   • [claimed-docs] “Logseq: A privacy-first, open-source knowledge base”
   • [community] “I use Logseq personally and for work almost daily since a year ago or so. It's pretty nice and the 'least bad' PKM platform I'm aware of (no…”

   proof ↗⚑ flag

   Reflect—none0/10

   The axis applies to this product kind (peer products hold positive or none verdicts on this story), so lack of evidence for an applicable capability is "none", never "na". (na/none harmonized at arena bring-up — see pipeline/scripts/na-harmonize.ts.)

   ⚑ flag

4. ai-native userSelf-host the core product

   weight 3 · round to Logseq
   Logseq~partialcommunity5/10

   Logseq is open-source and stores data locally as markdown files (docs-23), and offers a local HTTP API for accessing your own graph (logseq-docs-6, logseq-docs-21), which supports a 'run entirely on your own machine' notion of self-hosting. However, there is no official self-hosted server/web deployment: a community member explicitly wished they could self-host Logseq as a website with notes on their own server, noting this remains an open, unresolved GitHub feature request among 1.8k open issues (logseq-comm-20), and official Sync is a paid, vendor-hosted cloud service rather than a self-hostable component (logseq-docs-10, logseq-comm-2). Missing for 10: an official self-hosted server/backend deployment option, first-party documentation for hosting the app as a service, and confirmation that the open feature request has been resolved.

   • [claimed-docs] “Logseq: A privacy-first, open-source knowledge base”
   • [claimed-docs] “local HTTP server to access logseq's graph ... Enable HTTP APIs server in `Features > HTTP APIs server`”
   • [claimed-docs] “we developed the local HTTP server to support external access to the database through authorization.”
   • [community] “I know Logseq is intended as an offline-first solution, but I really wish I could self-host it as a website and have all notes saved on my s…”
   • [claimed-docs] “Syncs up to 10 graphs between Logseq Desktop, Android, and iOS.”
   • [community] “Logseq's own sync is now in testing and you can access it if you're a sponsor ($15/month tier)... It works fine but has enough bugs that I w…”

   proof ↗⚑ flag

   Reflect—none0/10

   The axis applies to this product kind (peer products hold positive or none verdicts on this story), so lack of evidence for an applicable capability is "none", never "na". (na/none harmonized at arena bring-up — see pipeline/scripts/na-harmonize.ts.)

   ⚑ flag

Metadata

1. power-userAttach typed properties and tags to notes and use them consistently for organization and retrieval

   weight 2 · round to Logseq
   Logseq~partialclaimed6/10

   Docs confirm block/page tagging (`#Page`) and properties (e.g., `publish:: true`), plus page/block references and query blocks (`{{query}}`) that can retrieve content by tags/properties. However, the evidence pack lacks dedicated documentation on Logseq's typed-property system (property schemas, value types, advanced query filtering by property) and no independent confirmation of consistent power-user workflows using properties for retrieval. Missing for 10: explicit docs on property types/schema, advanced query examples filtering by properties, and hands-on validation of consistent tagging/property retrieval at scale.

   • [claimed-docs] “Type `[[]]` to create or reference any other page. You can also make a new page in the search bar. - Type `(())` to reference a block on any…”
   • [claimed-docs] “Simple queries by using `{{query }}`, the format is something like this: {{query Something you're looking for}}”
   • [claimed-docs] “By default no pages are public. Individual pages can be marked as public with the property `publish:: true`.”
   • [claimed-docs] “Simple queries by using `{{query }}`... You can type `/query` to create a simple query.”
   • [claimed-docs] “Blocks that have a page as a parent i.e. usually top-level blocks can be converted to a page and back to a block easily. To convert this blo…”
   • [claimed-docs] “Queries are for asking questions from your knowledge base and the outside world (in the future).”

   proof ↗⚑ flag

   Reflect~partialclaimed5/10

   Reflect clearly supports tags (#tags) for organization and retrieval, including automatic tagging (e.g. Kindle imports tagged #book) and filtering notes by #tags, backlinks, dates, and pinned status. However, there is no evidence of 'typed properties' (structured metadata fields with types, akin to database properties) — Reflect's organization model is built around tags, backlinks, and daily notes, not typed property schemas. Missing for 10: evidence of typed/structured properties (not just tags), evidence of consistent property schemas across notes, and any UI for defining property types.

   • [claimed-docs] “You can filter notes by: Date created or updated, Daily Notes, Where they are backlinked to/from, #tags, Pinned notes”
   • [claimed-docs] “You can filter notes by: Date created or updated Daily Notes Where they are backlinked to/from #tags Pinned notes”
   • [claimed-docs] “You can filter notes by: * Date created or updated * Daily Notes * Where they are backlinked to/from * #tags * Pinned notes”
   • [claimed-docs] “Reflect can automatically sync your books, highlights, and comments from your Kindle. They will be imported into Reflect as individual notes…”
   • [claimed-docs] “Reflect can automatically sync your books, highlights, and comments from your Kindle. They will be imported into Reflect as individual notes…”

   proof ↗⚑ flag

Queries

1. power-userQuery my notes by their structured properties and render live database-style views (tables, boards, lists) from those queries

   weight 3 · round to Logseq
   Logseq~partialclaimed3/10

   Docs confirm a core query feature ({{query}} / /query) for asking questions of the knowledge base, and structured page/block references exist, but the pack never documents property-based structured queries or dynamic table/board/list rendering views (advanced query mode, query views like {{table}}) that power users need. missing for 10: documentation of advanced/structured-property queries, table/board/list view rendering from queries, and any hands-on confirmation of these views working.

   • [claimed-docs] “Simple queries by using `{{query }}`, the format is something like this: {{query Something you're looking for}}”
   • [claimed-docs] “Simple queries by using `{{query }}`... You can type `/query` to create a simple query.”
   • [claimed-docs] “Queries are for asking questions from your knowledge base and the outside world (in the future).”
   • [claimed-docs] “Blocks that have a page as a parent i.e. usually top-level blocks can be converted to a page and back to a block easily. To convert this blo…”

   proof ↗⚑ flag

   Reflect—none0/10

   Reflect only supports basic filtering (tags, dates, backlinks, pinned) per reflect-docs-15/22/27, with no evidence of structured/typed properties, query language, or database-style table/board/list views like Notion or Roam-style databases. No documentation or community evidence mentions live queryable views rendered from structured properties.

   • [claimed-docs] “You can filter notes by: Date created or updated, Daily Notes, Where they are backlinked to/from, #tags, Pinned notes”
   • [claimed-docs] “You can filter notes by: Date created or updated Daily Notes Where they are backlinked to/from #tags Pinned notes”
   • [claimed-docs] “You can filter notes by: * Date created or updated * Daily Notes * Where they are backlinked to/from * #tags * Pinned notes”

   proof ↗⚑ flag

Search

1. researcherInstantly search all my notes with full-text search plus filters on tags, properties, and paths — including offline

   weight 3 · round to Reflect
   Logseq!disputedcontradicted5/10

   Logseq's docs describe a query system ({{query}}) for searching the knowledge base and page/block references and tags for organization (logseq-docs-4, -16, -19, -3), and it markets itself as privacy-first/local/offline (logseq-docs-23). However, hands-on community reports directly contradict the 'instant' part of the story: users report that 'indexing and general responsiveness slows down massively once you have a few million words stored' and describe persistent lag navigating pages even after months of daily use (logseq-comm-5, logseq-comm-6). Missing for 10: explicit documentation of a dedicated full-text search UI with tag/property/path filters (only generic query syntax is shown), and no independent benchmark confirming fast search at scale to offset the reported slowdowns.

   • [claimed-docs] “Simple queries by using `{{query }}`, the format is something like this: {{query Something you're looking for}}”
   • [claimed-docs] “Simple queries by using `{{query }}`... You can type `/query` to create a simple query.”
   • [claimed-docs] “Queries are for asking questions from your knowledge base and the outside world (in the future).”
   • [claimed-docs] “Type `[[]]` to create or reference any other page. You can also make a new page in the search bar. - Type `(())` to reference a block on any…”
   • [claimed-docs] “Logseq: A privacy-first, open-source knowledge base”
   • [community] “Admittedly poor performance has been my primary gripe in the 7mos I've been using LogSeq on a near-daily basis... I'd gladly pay for a subsc…”
   • [community] “Indexing and general responsiveness slows down massively once you have a few million words stored, and you have to break up your data into m…”

   proof ↗⚑ flag

   Reflect~partialclaimed6/10

   Reflect documents semantic + full-text search with filters on tags, backlinks, date, daily notes, and pinned status (reflect-docs-6,15,22,27,41), covering most of the story. However, no explicit filter on "properties" (beyond tags) or "paths" is documented, and offline capability/search-while-offline is never mentioned anywhere in the evidence. Missing for 10: explicit property-based filtering, path-based filtering, and any documentation of offline search support.

   • [claimed-docs] “Reflect’s Semantic search lets you find specific pieces of information without knowing the exact words.”
   • [claimed-docs] “You can filter notes by: Date created or updated, Daily Notes, Where they are backlinked to/from, #tags, Pinned notes”
   • [claimed-docs] “You can filter notes by: Date created or updated Daily Notes Where they are backlinked to/from #tags Pinned notes”
   • [claimed-docs] “You can filter notes by: * Date created or updated * Daily Notes * Where they are backlinked to/from * #tags * Pinned notes”
   • [claimed-docs] “Reflect's Semantic search lets you find specific pieces of information without knowing the exact words.”
   • [claimed-docs] “After applying any desired filters to the search, you can also “chat” with those specific notes to ask questions and extract information.”

   proof ↗⚑ flag

Automation hooks

1. developerDrive the app from outside — create or open notes via a URI scheme, deep links, or a local endpoint — to wire it into my scripts and workflows

   weight 2 · round drawn
   Logseq✓fullprobed8/10

   Logseq documents both a URI scheme (logseq://x-callback-url/quickCapture) that can be triggered from browser bookmarks or command line to capture/open notes, and a local HTTP APIs server enabling authorized external programmatic access to the graph — directly matching the story's ask for deep links and a local endpoint for scripting. Missing for 10: independent/hands-on confirmation of the HTTP API's request/response schema or third-party integration examples, and no evidence of a broader public REST API spec (openapi probes returned 404).

   • [claimed-docs] “local HTTP server to access logseq's graph ... Enable HTTP APIs server in `Features > HTTP APIs server`”
   • [claimed-docs] “logseq://x-callback-url/quickCapture? ... Quick capture for short text and links. Can be added as a bookmark link to modern browser”
   • [claimed-docs] “logseq://x-callback-url/quickCapture? - Quick capture for short text and links.”
   • [claimed-docs] “we developed the local HTTP server to support external access to the database through authorization.”
   • [claimed-docs] “Logseq provides a protocol `logseq://` to allow third party applications to invoke actions within Logseq.”
   • [claimed-docs] “Quick capture for short text and links. Can be added as a bookmark link to modern browser or opened from command line shell”
   • [probe] “PROBE openapi: all candidate paths 404 (https://plugins-doc.logseq.com/openapi.json, https://plugins-doc.logseq.com/swagger.json, https://pl…”

   proof ↗⚑ flag

   Reflect✓fullprobed8/10

   Reflect ships a documented custom URI scheme (reflect://reflect?command=...) for deep linking into the app, plus a REST API with OAuth2/PKCE authentication that a runtime probe confirms is live and enforcing auth, letting developers script note creation/appending from outside the app. missing for 10: independent/hands-on developer confirmation of building an external integration, and full endpoint/command reference beyond the docs snippets.

   • [claimed-docs] “Reflect supports deep linking, allowing users to perform actions within the app using custom URLs.”
   • [claimed-docs] “reflect://reflect?command=<command_name>&<parameter1>=<value1>&...”
   • [claimed-docs] “For authentication, we support OAuth 2. You will need to create credentials here.”
   • [claimed-docs] “Our supports appending data to your notes, returning a list of the links you've bookmarked, and a few other things.”
   • [claimed-docs] “We do support OAuth's PKCE extension which means you can build a client to Reflect without shipping secret keys around.”
   • [probe] “PROBE runtime (recorded 2026-09-05): keyless GET to the documented REST base https://reflect.app/api/graphs returned HTTP 401 {"error":{"typ…”

   proof ↗⚑ flag

Plugins

1. power-userBrowse and install from a large catalog of community plugins/extensions to add features the core app doesn't ship

   weight 3 · round to Logseq
   Logseq✓fullprobed8/10

   Logseq ships an in-app Marketplace dashboard to search/install/update/remove third-party plugins, plus a documented plugin API and community docs site, and a runtime probe confirms 617 published plugin packages in the official registry. Community commentary notes Logseq's ecosystem is smaller/less mature than Obsidian's, but doesn't dispute that a working plugin catalog exists. Missing for 10: independent hands-on account of browsing/installing a plugin, and quantification of catalog quality/curation vs. just count.

   • [claimed-docs] “Extends the functionality of logseq with third party tools ... third-party plugins from the Marketplace, can be searched, installed, updated…”
   • [claimed-docs] “From this dashboard, third-party plugins from the [[Marketplace]], can be searched, installed, updated and removed.”
   • [claimed-docs] “Extends the functionality of logseq with third party tools ... third-party plugins from the [[Marketplace]], can be searched, installed, upd…”
   • [claimed-docs] “Click on `Load unpacked plugin` to load plugins that are distributed outside the marketplace.”
   • [claimed-docs] “third-party plugins from the [[Marketplace]], can be searched, installed, updated and removed.”
   • [claimed-docs] “This is a community-driven documentation for Logseq plugin APIs.”
   • [probe] “PROBE runtime (recorded 2026-09-05): Logseq's official plugin marketplace registry (logseq/marketplace, one directory per package) counted 6…”
   • [community] “Yeah I felt the same about the decision when they decided to move to database structure. If it had this structure from the start it could've…”

   proof ↗⚑ flag

   Reflect—none0/10

   There is no evidence of a plugin/extension marketplace; instead community evidence explicitly states Reflect is designed to be 'fully functional without any plugins' as a differentiator from Obsidian, indicating no such catalog exists.

   • [community] “What drew me to Reflect (away from Obsidian) was full functionality out of the box without the need for plugins... Reflect is fully function…”

   proof ↗⚑ flag

2. developerBuild my own plugins against a documented extension API/SDK with samples and developer docs

   weight 2 · round to Logseq
   Logseq~partialprobed6/10

   Logseq has a documented plugin API (plugins-doc.logseq.com, described as community-driven documentation for the SDK) plus in-app tooling for loading unpacked/dev plugins and a marketplace with 617 published packages showing a real developer ecosystem. However, the core API docs are explicitly community-maintained rather than official first-party reference docs with samples, and probes show no llms.txt/openapi spec, suggesting thinner formal SDK documentation than top competitors. Missing for 10: official first-party API reference with code samples/tutorials, and machine-readable API spec (openapi/llms.txt) confirming documented completeness.

   • [claimed-docs] “Extends the functionality of logseq with third party tools ... third-party plugins from the Marketplace, can be searched, installed, updated…”
   • [claimed-docs] “This is a community-driven documentation for Logseq plugin APIs.”
   • [claimed-docs] “Click on `Load unpacked plugin` to load plugins that are distributed outside the marketplace.”
   • [probe] “PROBE runtime (recorded 2026-09-05): Logseq's official plugin marketplace registry (logseq/marketplace, one directory per package) counted 6…”
   • [probe] “PROBE llms.txt: HTTP 404 at https://plugins-doc.logseq.com/llms.txt”
   • [probe] “PROBE openapi: all candidate paths 404 (https://plugins-doc.logseq.com/openapi.json, https://plugins-doc.logseq.com/swagger.json, https://pl…”

   proof ↗⚑ flag

   Reflect—none0/10

   Reflect documents a limited OAuth2 REST API for appending data/links and deep-linking (reflect-docs-2,17,18,38), but this is not a plugin extension SDK, and community evidence explicitly states Reflect is 'fully functional without any plugins' by design, unlike Obsidian (reflect-comm-2). No plugin architecture, extension samples, or plugin developer docs exist.

   • [claimed-docs] “Our supports appending data to your notes, returning a list of the links you've bookmarked, and a few other things.”
   • [claimed-docs] “We do support OAuth's PKCE extension which means you can build a client to Reflect without shipping secret keys around.”
   • [community] “What drew me to Reflect (away from Obsidian) was full functionality out of the box without the need for plugins... Reflect is fully function…”

   proof ↗⚑ flag

1. ai-native userChoose where my data is stored (region/residency)

   weight 2 · round to Logseq
   Logseq~partialcommunity6/10

   Logseq stores data as local markdown/EDN files by default, giving users full control over where their data physically resides (their own device, self-hosted git repos, etc.), and its optional Sync service is end-to-end encrypted, which aligns with a privacy-first, local-first posture. However, there is no documented feature letting users select a specific data-center region for the Sync service, and community reports note sync reliability issues and no self-hosting option for a hosted graph. Missing for 10: explicit region/residency selection controls for the cloud Sync product, first-party documentation on where Sync servers are physically located, and confirmation that self-hosting/regional choice is officially supported rather than just achievable via local storage or third-party git.

   • [claimed-docs] “Logseq: A privacy-first, open-source knowledge base”
   • [claimed-docs] “Syncs up to 10 graphs between Logseq Desktop, Android, and iOS.”
   • [claimed-docs] “Yes, your graph is end-to-end encrypted.”
   • [community] “I switched over to LogSeq from Roam Research, and I'm using Github for syncing and backup.”
   • [community] “Logseq's own sync is now in testing and you can access it if you're a sponsor ($15/month tier)... It works fine but has enough bugs that I w…”
   • [community] “I know Logseq is intended as an offline-first solution, but I really wish I could self-host it as a website and have all notes saved on my s…”

   proof ↗⚑ flag

   Reflect—none0/10

   No evidence anywhere in the pack mentions data residency, region selection, or storage location choices; Reflect discusses E2E encryption and export options but nothing about choosing where data is stored geographically.

   ⚑ flag

2. ai-native userPrevent my data from being used to train AI models

   weight 3 · round to Logseq
   Logseq~partialcommunity3/10

   Logseq positions itself as a 'privacy-first, open-source knowledge base' with local-first storage and end-to-end encrypted sync, which implies user data isn't funneled to a cloud AI training pipeline by default, but no evidence explicitly states a policy on AI model training use of user data. Missing for 10: explicit statement addressing AI training data usage, any opt-out/consent controls for AI features, and confirmation that plugins or integrations (e.g., Claude/Codex mentioned in community posts) don't send data to third-party AI services.

   • [claimed-docs] “Logseq: A privacy-first, open-source knowledge base”
   • [claimed-docs] “Yes, your graph is end-to-end encrypted.”
   • [community] “I've got so much already in logseq though, and I really like the model. Right now I'm thinking I'll just stay on version one as long as poss…”

   proof ↗⚑ flag

   Reflect—none0/10

   Reflect advertises end-to-end encryption of note content (reflect-docs-16, reflect-docs-35), which limits Reflect's own server-side access, but there is no explicit statement anywhere in the evidence about whether note content sent to third-party AI models (GPT-4o, Claude 3.5 Sonnet per reflect-docs-37) is used for model training, nor any opt-out/data-processing policy addressing AI training use. Community evidence even notes files/images are excluded from E2E encryption (reflect-comm-8), raising unaddressed questions rather than resolving them.

   • [claimed-docs] “End-to-end encryption keeps your notes secure.”
   • [claimed-docs] “The contents of your notes are end-to-end encrypted. No one else can read them (not even us).”
   • [claimed-docs] “Reflect's AI tools give you the option to now use either GPT-4o (the latest model from OpenAI) or Claude 3.5 Sonnet (the latest model from A…”
   • [community] “Question about Reflect's E2E encryption claims: 'We do not end-to-end encrypt things like files or images you upload' - user noted this excl…”

   proof ↗⚑ flag

3. ai-native userControl data retention and deletion

   weight 2 · round to Logseq
   Logseq~partialclaimed5/10

   Logseq stores graphs as local Markdown files and is marketed as 'privacy-first,' implying users can delete or retain data simply by managing local files, and Export lets users extract data in multiple formats; optional Sync is end-to-end encrypted. However, there is no explicit documentation on deletion mechanics for synced/cloud data, data-retention policies, or how to purge synced copies across devices. Missing for 10: documented data-retention/deletion policy, explicit instructions for deleting synced cloud data, and independent verification that deletion is complete and irreversible.

   • [claimed-docs] “Logseq: A privacy-first, open-source knowledge base”
   • [claimed-docs] “The Logseq Export feature allows users to export their data in various formats, including text, HTML, OPML, PNG, EDN, JSON, and standard Mar…”
   • [claimed-docs] “Syncs up to 10 graphs between Logseq Desktop, Android, and iOS.”
   • [claimed-docs] “Yes, your graph is end-to-end encrypted.”

   proof ↗⚑ flag

   Reflect—none0/10

   Evidence covers export/backup, encryption, and API access, but there is no documentation or mention of account/data deletion, retention settings, or right-to-erasure controls anywhere in the pack.

   ⚑ flag

4. ai-native userOpt out of telemetry and usage tracking

   weight 2 · round drawn
   Logseq—none0/10

   The evidence describes Logseq as 'privacy-first, open-source' and local-first, implying no telemetry infrastructure exists, but there is no explicit documentation of a telemetry/usage-tracking feature or an opt-out setting for it. Absence of concrete evidence of an actual telemetry toggle means this cannot be confirmed as full/partial delivery. Missing for 10: explicit telemetry/analytics policy documentation, evidence of an opt-out setting or confirmation that no tracking occurs at all.

   • [claimed-docs] “Logseq: A privacy-first, open-source knowledge base”

   proof ↗⚑ flag

   Reflect—none0/10

   No evidence pack item discusses telemetry, analytics, usage tracking, or an opt-out/privacy settings toggle for such tracking; the evidence covers encryption, imports, AI features, and API access instead. Missing for 10: any mention of telemetry collection, opt-out settings, or a privacy policy addressing usage tracking.

   ⚑ flag

Publish

1. researcherPublish selected notes as a browsable website — a digital garden with working links — without leaving the app

   weight 2 · round to Logseq
   Logseq✓fullclaimed7/10

   Logseq's Publishing docs confirm per-page publish:: true flags to selectively expose notes, and the Publish feature generates a browsable static SPA site with working page/block references intact, matching the digital-garden use case entirely within the app. missing for 10: no independent/hands-on report confirming the exported site's link integrity or ease of workflow, and no detail on hosting/deployment specifics or custom domain support.

   • [claimed-docs] “By default no pages are public. Individual pages can be marked as public with the property `publish:: true`.”
   • [claimed-docs] “This feature publishes a graph as a [[Publish Web]], single page application (SPA).”
   • [claimed-docs] “Type `[[]]` to create or reference any other page. You can also make a new page in the search bar. - Type `(())` to reference a block on any…”

   proof ↗⚑ flag

   Reflect~partialclaimed5/10

   Reflect docs confirm one-click publishing of individual notes to a public URL with minimal branding, which supports sharing selected notes without leaving the app, but there is no evidence of a browsable multi-page 'digital garden' site (index/navigation) or confirmation that backlinks between published notes remain functional as working links. Missing for 10: evidence of a site-level index/navigation across multiple published notes, confirmation that internal links/backlinks resolve correctly between published pages, and any custom-domain or site-branding options.

   • [claimed-docs] “Reflect offers one-click publishing of your notes. Simply hit the publish button and your note will be publicly available behind a long hard…”
   • [claimed-docs] “We try to keep the style clean and branding to a minimum.”
   • [claimed-docs] “Backlinks are just a list of incoming links from other notes.”

   proof ↗⚑ flag

Collaboration

1. researcherShare a vault or space with collaborators and edit notes together without stepping on each other's changes

   weight 2 · round drawn
   Logseq—none0/10

   Logseq's Sync feature is described as syncing a user's own graphs across their own devices (desktop/Android/iOS), not as multi-user real-time collaborative editing with conflict resolution; community reports describe using GitHub for personal sync/backup and note Logseq's own sync was still buggy/beta, with no evidence of concurrent multi-collaborator editing safeguards.

   • [claimed-docs] “Syncs up to 10 graphs between Logseq Desktop, Android, and iOS.”
   • [claimed-docs] “Yes, your graph is end-to-end encrypted.”
   • [community] “I switched over to LogSeq from Roam Research, and I'm using Github for syncing and backup.”
   • [community] “Logseq's own sync is now in testing and you can access it if you're a sponsor ($15/month tier)... It works fine but has enough bugs that I w…”
   • [community] “I still use Logseq and conceptually it's still a great method for building a second brain. But it has been dormant for years and early attem…”

   proof ↗⚑ flag

   Reflect—none0/10

   No evidence pack item mentions shared vaults/spaces, multi-user collaboration, or real-time co-editing/conflict resolution; Reflect's documented features are import/export, linking, AI, search, publishing, integrations, and E2E encryption for a single user's notes, with no mention of collaborative editing between multiple people.

   ⚑ flag

Sync

1. power-userAn official sync service keeps my notes updated across all my devices with end-to-end encryption the vendor can't read

   weight 3 · round drawn
   Logseq!disputedcontradicted5/10

   Docs confirm an official Logseq Sync service across Desktop/Android/iOS with explicit end-to-end encryption ('Yes, your graph is end-to-end encrypted'), fitting the story's claim. However, community reports directly contradict reliable delivery: one user in 2022 found the paid sync beta 'has enough bugs that I wouldn't rely on it yet,' and another in 2025 stated 'early attempts at syncing didn't work well... I paid to support the sync effort but we saw nothing for years,' indicating the sync feature has a documented history of not working as promised despite payment. Missing for 10: independent verification that sync is now stable/production-ready, technical detail on the E2EE key model, and resolution of the multi-year reliability complaints.

   • [claimed-docs] “Syncs up to 10 graphs between Logseq Desktop, Android, and iOS.”
   • [claimed-docs] “Yes, your graph is end-to-end encrypted.”
   • [community] “Logseq's own sync is now in testing and you can access it if you're a sponsor ($15/month tier)... It works fine but has enough bugs that I w…”
   • [community] “I still use Logseq and conceptually it's still a great method for building a second brain. But it has been dormant for years and early attem…”
   • [community] “I loved what Logseq had to offer, and setup monthly donations for a while, but after the long period of apparent stagnation I lost faith and…”

   proof ↗⚑ flag

   Reflect!disputedcontradicted5/10

   Reflect markets end-to-end encryption 'no one else can read them (not even us)' and syncs notes across devices as a core feature, but a founder-sourced community comment reveals a concrete carve-out: files/images uploaded are NOT end-to-end encrypted, contradicting the blanket vendor claim. Missing for 10: explicit documentation of sync architecture/conflict resolution across devices, independent security audit report details, and a full accounting of what content types are/aren't encrypted.

   • [claimed-docs] “End-to-end encryption keeps your notes secure.”
   • [claimed-docs] “The contents of your notes are end-to-end encrypted. No one else can read them (not even us).”
   • [community] “Question about Reflect's E2E encryption claims: 'We do not end-to-end encrypt things like files or images you upload' - user noted this excl…”
   • [community] “Reflect founder: 'We're building it as a lifestyle company. There's only four of us. We bootstrapped for a while, and then did a $1m crowdfu…”
   • [community] “Props to Reflect for their security audit too. I will be keeping an eye on how they progress.”

   proof ↗⚑ flag

2. power-userView a note's version history and restore earlier versions or recover deleted notes

   weight 2 · round to Reflect
   Logseq—none0/10

   The evidence pack contains no documentation of a native version-history or restore/undo-delete feature in Logseq; users instead rely on external tools like Git for backup (logseq-comm-1), and there are reports of past data-loss issues (logseq-comm-3), but no built-in mechanism to view or restore prior versions of a note is described.

   • [community] “I switched over to LogSeq from Roam Research, and I'm using Github for syncing and backup.”
   • [community] “I had two main problems with LogSeq: 1. Not designed from the ground-up to prevent data loss. Has had data-loss issues in the past.”

   proof ↗⚑ flag

   Reflect✓fullclaimed8/10

   Reflect's documentation explicitly states it continually records edits and lets users revert a note to any point in time, directly matching the version history/restore story (docs-8). Deleted note recovery isn't explicitly documented, so missing for 10: explicit documentation of deleted-note recovery/trash restoration, and independent hands-on corroboration of restore functionality.

   • [claimed-docs] “We continually record edits to notes which means you can see all your changes and revert a note to any point in time.”

   proof ↗⚑ flag