Rank #1 of 5 in Agent Skills & Extensions
Access
vs Superpowers
vs Skills for Real Engineers
vs skills.sh
vs Codex PluginsShowcase
Products
Anthropic, product by product →Anthropic ships more than one product — each judged line competes in its own arena on the same stories as everyone else.
| Line | Arena | Rank | PA Score | Agent-ready |
|---|---|---|---|---|
| Claude | AI Assistants | #2/9 | 29/100 | 28/100 |
| Claude Code | AI Coding Agents | #2/13 | 40/100 | 69/100 |
| Claude Design | Design & Prototyping | #8/8 | 18/100 | 16/100 |
| Claude Agent SDK | Agent Frameworks & SDKs | #1/9 | 38/100 | 67/100 |
| Anthropic Skillsthis page | Agent Skills & Extensions | #1/5 | 24/100 | 54/100 |
Not yet judged (8 — no arena where they compete): Claude Cowork · Claude in Chrome · @Claude (Slack & Teams) · Claude for Microsoft 365 · Claude Science · Claude Security · Managed Agents · Claude Developer Platform
Try itExperimental
See what an agent can do with Anthropic Skills before you ever sign up. Pick a story: recorded sessions replay real probe-harness transcripts; sandboxed self-drive sessions are designed and gated (docs/TRY-IT.md).
$npx -y skills add anthropics/skills --skill skill-creator -a claude-code -y --copy # in a scratch dir, then print installed SKILL.md frontmatterrecorded session — replayed, not liveVerified integrations
Connections to other tracked products — hover a chip for the verbatim evidence quote behind it.
By theme — the product's score on each story themeBy theme
Agent workflows — stories about agent workflows in this arenaAgent workflowsevidence →
Stories about agent workflows in this arena
Agenticness — how well agents can access and operate the productAgenticnessevidence →
How well agents can access and operate the product
Automation depth — how much of the product can run unattendedAutomation depthevidence →
How much of the product can run unattended
Cross agent portability — stories about cross agent portability in this arenaCross agent portabilityevidence →
Stories about cross agent portability in this arena
Discovery distribution — stories about discovery distribution in this arenaDiscovery distributionevidence →
Stories about discovery distribution in this arena
Docs onboarding — stories about docs onboarding in this arenaDocs onboardingevidence →
Stories about docs onboarding in this arena
Install experience — stories about install experience in this arenaInstall experienceevidence →
Stories about install experience in this arena
Openness — open source, data portability, and self-hosting storiesOpennessevidence →
Open source, data portability, and self-hosting stories
Privacy posture — data-handling and privacy storiesPrivacy postureevidence →
Data-handling and privacy stories
Safety review — stories about safety review in this arenaSafety reviewevidence →
Stories about safety review in this arena
Skill authoring — stories about skill authoring in this arenaSkill authoringevidence →
Stories about skill authoring in this arena
Testing quality — stories about testing quality in this arenaTesting qualityevidence →
Stories about testing quality in this arena
Versioning updates — stories about versioning updates in this arenaVersioning updatesevidence →
Stories about versioning updates in this arena
Story verdicts — every judged story with its evidenceStory verdicts
What’s free: 0 free · 1 paid · 0 enterprise · 26 not stated in evidence
Follow the green: where the map greys out is where Anthropic Skills stops today. ✓ full · ~ partial · ! disputed · — none · n/a not applicable.
Agent workflows — stories about agent workflows in this arenaAgent workflows
Stories about agent workflows in this arena
Agenticness — how well agents can access and operate the productAgenticness
How well agents can access and operate the product
API surface
Drive the product through a documented public API
✓8/10
unlocks → Versioning policy · Full data export
Subscribe to events via webhooks
n/an/a
Build against official SDKs
✓7/10
Issue scoped/least-privilege API credentials for an agent
n/an/a
Connect an agent via an official MCP server
n/an/a
Download a machine-readable API spec (OpenAPI or equivalent)
~3/10
Rely on versioned APIs with a documented deprecation policy
—–
Test against a sandbox environment without touching production data
~4/10
Explore an interactive API reference with runnable examples
n/an/a
Docs for agents
Point an agent at llms.txt or agent-oriented docs
✓8/10
Agentic features
Delegate tasks to a built-in AI assistant inside the product
!5/10
Operate the product with natural-language commands
!5/10
Plug MCP servers into this product so it can use their tools
~5/10
Get AI-generated insights and suggestions from my data inside the product
~4/10
Set up automations that run autonomously in the background
—0/10
Automation depth — how much of the product can run unattendedAutomation depth
How much of the product can run unattended
Cross agent portability — stories about cross agent portability in this arenaCross agent portability
Stories about cross agent portability in this arena
Discovery distribution — stories about discovery distribution in this arenaDiscovery distribution
Stories about discovery distribution in this arena
Browse or search a catalog of available skills — a registry, leaderboard, or marketplace listing — before installing anything
~6/10
Distribute a standard skill set to my whole team — via a marketplace, a shared repo, or files committed to the project
✓8/10
Installed skills trigger automatically from task context, with descriptions engineered so the agent activates the right skill at the right moment
!5/10
Docs onboarding — stories about docs onboarding in this arenaDocs onboarding
Stories about docs onboarding in this arena
Install experience — stories about install experience in this arenaInstall experience
Stories about install experience in this arena
Install
List what is installed and remove skills cleanly, without orphaned files or lingering instructions
—0/10
Openness — open source, data portability, and self-hosting storiesOpenness
Open source, data portability, and self-hosting stories
Privacy posture — data-handling and privacy storiesPrivacy posture
Data-handling and privacy stories
Safety review — stories about safety review in this arenaSafety review
Stories about safety review in this arena
Skill authoring — stories about skill authoring in this arenaSkill authoring
Stories about skill authoring in this arena
Testing quality — stories about testing quality in this arenaTesting quality
Stories about testing quality in this arena
Versioning updates — stories about versioning updates in this arenaVersioning updates
Stories about versioning updates in this arena
Sorted by importance (agentic first) (high → low) · 52/52 stories · click a row’s chevron for the rationale and evidence
Drive the product through a documented public API G Agent access | ai-native user | Agenticness — how well agents can access and operate the productAgenticness | 3 | full | 8/10 | Cclaimed | |
Delegate tasks to a built-in AI assistant inside the product G Agentic features | ai-native user | Agenticness — how well agents can access and operate the productAgenticness | 3 | disputed | 5/10 | Dcontradicted | |
Plug MCP servers into this product so it can use their tools G Agent access | ai-native user | Agenticness — how well agents can access and operate the productAgenticness | 3 | partial | 5/10 | Xcommunity | |
Connect an agent via an official MCP server G Agent access | ai-native user | Agenticness — how well agents can access and operate the productAgenticness | 3 | n/a | untested | none yet | |
Point an agent at llms.txt or agent-oriented docs G Agent access | ai-native user | Agenticness — how well agents can access and operate the productAgenticness | 2 | full | 8/10 | Tprobed | |
Build against official SDKs G Agent access | ai-native user | Agenticness — how well agents can access and operate the productAgenticness | 2 | full | 7/10 | Cclaimed | |
Operate the product with natural-language commands G Agentic features | ai-native user | Agenticness — how well agents can access and operate the productAgenticness | 2 | disputed | 5/10 | Dcontradicted | |
Run the product headlessly / in CI for automation G Agent access | ai-native user | Agenticness — how well agents can access and operate the productAgenticness | 2 | partial | 5/10 | Cclaimed | |
Use an official CLI G Agent access | ai-native user | Agenticness — how well agents can access and operate the productAgenticness | 2 | partial | 5/10 | Tprobed | |
Get AI-generated insights and suggestions from my data inside the product G Agentic features | ai-native user | Agenticness — how well agents can access and operate the productAgenticness | 2 | partial | 4/10 | Xcommunity | |
Download a machine-readable API spec (OpenAPI or equivalent) G Api quality | ai-native user | Agenticness — how well agents can access and operate the productAgenticness | 2 | partial | 3/10 | Cclaimed | |
Set up automations that run autonomously in the background G Agentic features | ai-native user | Agenticness — how well agents can access and operate the productAgenticness | 2 | none | 0/10 | ||
Explore an interactive API reference with runnable examples G Api quality | ai-native user | Agenticness — how well agents can access and operate the productAgenticness | 2 | n/a | untested | none yet | |
Issue scoped/least-privilege API credentials for an agent G Agent access | ai-native user | Agenticness — how well agents can access and operate the productAgenticness | 2 | n/a | untested | none yet | |
Rely on versioned APIs with a documented deprecation policy G Api quality | ai-native user | Agenticness — how well agents can access and operate the productAgenticness | 2 | none | untested | none yet | |
Subscribe to events via webhooks G Agent access | ai-native user | Agenticness — how well agents can access and operate the productAgenticness | 2 | n/a | untested | none yet | |
Test against a sandbox environment without touching production data G Api quality | ai-native user | Agenticness — how well agents can access and operate the productAgenticness | 1 | partial | 4/10 | Cclaimed | |
Author a new skill from a documented template — a SKILL.md with name and description frontmatter — without reverse-engineering existing skills C Authoring | developer | Skill authoring — stories about skill authoring in this arenaSkill authoring | 3 | full | 8/10 | Xcommunity | |
A quickstart takes me from nothing to a working installed skill in under five minutes C Onboarding | developer | Docs onboarding — stories about docs onboarding in this arenaDocs onboarding | 3 | partial | 6/10 | Xcommunity | |
Install a skill collection with one documented command — a package-manager one-liner, CLI, or in-agent marketplace command — and it is active in my next session C Install | developer | Install experience — stories about install experience in this arenaInstall experience | 3 | partial | 6/10 | Cclaimed | |
Review exactly what instructions and scripts a skill will add — list contents before installing and read every file afterward C Review | engineering-lead | Safety review — stories about safety review in this arenaSafety review | 3 | partial | 6/10 | Cclaimed | |
There is a documented update path — marketplace auto-updates or an explicit update command — so I get fixes without reinstalling from scratch C Updates | developer | Versioning updates — stories about versioning updates in this arenaVersioning updates | 3 | partial | 6/10 | Cclaimed | |
Installed skills trigger automatically from task context, with descriptions engineered so the agent activates the right skill at the right moment C Triggering | developer | Discovery distribution — stories about discovery distribution in this arenaDiscovery distribution | 3 | disputed | 5/10 | Dcontradicted | |
Define rules that trigger actions automatically on events G | ai-native user | Automation depth — how much of the product can run unattendedAutomation depth | 3 | disputed | 4/10 | Dcontradicted | |
Install the same collection into multiple different coding agents — Claude Code, Codex, Cursor, and others — with per-harness instructions C Portability | developer | Cross agent portability — stories about cross agent portability in this arenaCross agent portability | 3 | partial | 4/10 | Cclaimed | |
My coding agent can install a skill by itself — a non-interactive, promptless install path an agent can run headlessly end to end C Agent ops | ai-native user | Agent workflows — stories about agent workflows in this arenaAgent workflows | 3 | partial | 4/10 | Cclaimed | |
Export all of my data in open formats and leave G | ai-native user | Openness — open source, data portability, and self-hosting storiesOpenness | 3 | none | untested | none yet | |
Prevent my data from being used to train AI models G | ai-native user | Privacy posture — data-handling and privacy storiesPrivacy posture | 3 | n/a | untested | none yet | |
Self-host the core product G | ai-native user | Openness — open source, data portability, and self-hosting storiesOpenness | 3 | n/a | untested | none yet | |
My agent can author and package a new skill end to end by following the project's own spec, template, or meta-skill C Agent ops | ai-native user | Agent workflows — stories about agent workflows in this arenaAgent workflows | 2 | full | 9/10 | Cclaimed | |
Skills are plain markdown files and folders I can read, copy, and carry to another harness — not a proprietary binary format C Portability | developer | Cross agent portability — stories about cross agent portability in this arenaCross agent portability | 2 | full | 9/10 | Cclaimed | |
Distribute a standard skill set to my whole team — via a marketplace, a shared repo, or files committed to the project C Distribution | engineering-lead | Discovery distribution — stories about discovery distribution in this arenaDiscovery distribution | 2 | full | 8/10 | Cclaimed | |
Skills follow the open Agent Skills specification so the same skill folder is valid beyond this one vendor's tooling C Spec | developer | Skill authoring — stories about skill authoring in this arenaSkill authoring | 2 | full | 8/10 | Cclaimed | |
The collection ships a meta-skill or tool that guides my agent through writing, improving, and packaging new skills C Authoring | developer | Skill authoring — stories about skill authoring in this arenaSkill authoring | 2 | full | 8/10 | Cclaimed | |
Browse or search a catalog of available skills — a registry, leaderboard, or marketplace listing — before installing anything C Discovery | developer | Discovery distribution — stories about discovery distribution in this arenaDiscovery distribution | 2 | partial | 6/10 | Cclaimed | |
Install only the specific skills I want from a collection instead of taking the whole bundle C Install | developer | Install experience — stories about install experience in this arenaInstall experience | 2 | partial | 6/10 | Cclaimed | |
Do everything through the API that I can do in the UI G | ai-native user | Openness — open source, data portability, and self-hosting storiesOpenness | 2 | partialpaid | 5/10 | Cclaimed | |
Every skill documents what it does and when it activates, so I can predict my agent's new behavior before it surprises me C Onboarding | developer | Docs onboarding — stories about docs onboarding in this arenaDocs onboarding | 2 | disputed | 5/10 | Dcontradicted | |
Choose install scope — project-local files committed with my repo, or user-global across all projects C Install | developer | Install experience — stories about install experience in this arenaInstall experience | 2 | partial | 4/10 | Cclaimed | |
Perform bulk operations across many items at once G | ai-native user | Automation depth — how much of the product can run unattendedAutomation depth | 2 | partial | 3/10 | Xcommunity | |
Schedule recurring jobs or workflows G | ai-native user | Automation depth — how much of the product can run unattendedAutomation depth | 2 | none | 0/10 | ||
The collection maintains tests or evals for its skills so changes are verified against regressions rather than shipped on vibes C Testing | developer | Testing quality — stories about testing quality in this arenaTesting quality | 2 | none | 0/10 | ||
Choose where my data is stored (region/residency) G | ai-native user | Privacy posture — data-handling and privacy storiesPrivacy posture | 2 | n/a | untested | none yet | |
Control data retention and deletion G | ai-native user | Privacy posture — data-handling and privacy storiesPrivacy posture | 2 | none | untested | none yet | |
Opt out of telemetry and usage tracking G | ai-native user | Privacy posture — data-handling and privacy storiesPrivacy posture | 2 | n/a | untested | none yet | |
Read the product's source under an open license G | ai-native user | Openness — open source, data portability, and self-hosting storiesOpenness | 2 | none | untested | none yet | |
The collection is actively maintained — recent releases, triaged issues, and accepted community contributions C Maintenance | developer | Testing quality — stories about testing quality in this arenaTesting quality | 2 | none | untested | none yet | |
The project documents its security posture — what skills can execute, the trust model for third-party skills, and any telemetry or data collection C Trust | engineering-lead | Safety review — stories about safety review in this arenaSafety review | 2 | none | untested | none yet | |
Control when skill changes reach my team — pinned versions or a lockfile rather than silent behind-the-back updates C Pinning | engineering-lead | Versioning updates — stories about versioning updates in this arenaVersioning updates | 1 | partial | 5/10 | Cclaimed | |
Version, review, and roll back my automations G | ai-native user | Automation depth — how much of the product can run unattendedAutomation depth | 1 | partial | 4/10 | Cclaimed | |
List what is installed and remove skills cleanly, without orphaned files or lingering instructions C Lifecycle | developer | Install experience — stories about install experience in this arenaInstall experience | 1 | none | 0/10 | ||
Releases ship with notes or a changelog so I can see what changed in the skills before I take an update C Updates | developer | Versioning updates — stories about versioning updates in this arenaVersioning updates | 1 | none | untested | none yet |
Opportunities — the stories that would move this product's scores, from its own judged verdictsOpportunitiestop 8 of 30 stories with headroom
What would move Anthropic Skills’s scores — derived from its own judged verdicts, biggest headroom first. Each line quotes what the judge found missing; shipping it (or evidencing it publicly) is the fix.
Openness — open source, data portability, and self-hosting storiesExport all of my data in open formats and leave
nonemoves PA Scoreimpact 30
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".
Agenticness — how well agents can access and operate the productSet up automations that run autonomously in the background
nonemoves Built-in AIimpact 30
Skills are packaged instructions/capabilities that Claude loads and uses during a session (invoked automatically or via /skill-name), but the evidence pack contains no mention of scheduling, triggers, or background/autonomous execution outside an active user session.
Agenticness — how well agents can access and operate the productRely on versioned APIs with a documented deprecation policy
nonemoves API qualityimpact 30
Missing: any documentation of API version numbers, backward-compatibility guarantees, or deprecation timelines/policy.
Agenticness — how well agents can access and operate the productPlug MCP servers into this product so it can use their tools
partialq5/10moves agent-readyimpact 22.5
Missing: concrete first-party guide/example of installing an MCP server via a skill or plugin, hands-on confirmation that MCP tools become usable once added this way, and clarity on whether Skills (as opposed to Claude Code plugins broadly) directly expose MCP tool use.
Agenticness — how well agents can access and operate the productDownload a machine-readable API spec (OpenAPI or equivalent)
partialq3/10moves API qualityimpact 21
Missing: an actual OpenAPI/JSON-schema file for the Skills API endpoints, explicit download link/format, and independent confirmation that AI-native tooling consumes it.
Testing quality — stories about testing quality in this arenaThe collection is actively maintained — recent releases, triaged issues, and accepted community contributions
nonemoves PA Scoreimpact 20
Evidence pack covers docs, specification, and usage patterns for Skills, but contains no information about release cadence, issue triage, or acceptance of community contributions to the anthropics/skills repository.
Automation depth — how much of the product can run unattendedSchedule recurring jobs or workflows
nonemoves PA Scoreimpact 20
The evidence pack describes Skills as on-demand or auto-triggered instruction modules invoked by Claude during a task (via /skill-name, semantic triggering, or API container calls), but nothing describes a scheduler, cron-like trigger, or persistent recurring job mechanism.
Openness — open source, data portability, and self-hosting storiesRead the product's source under an open license
nonemoves PA Scoreimpact 20
While the anthropics/skills GitHub repo makes example skill files (SKILL.md, templates) publicly readable, none of the evidence cites an open-source license for Skills, Claude Code, or the underlying product; core Claude Code/Skills functionality itself is closed, proprietary tooling with no license grant shown.
Showing the top 8 of 30 — every none/partial verdict in the story verdicts table is headroom.
Think a verdict is wrong? Every verdicts-table row has a Flag link — see the methodology.
Coverage map — which docs area, API section, or community source covers which judged storiesCoverage map7 surfaces · 33 covered stories
Where the cited evidence behind each covered verdict came from — the same citations the verdicts table shows, no extra judging.
docs24 stories
- My agent can author and package a new skill end to end by following the project's own spec, template, or meta-skill
- My coding agent can install a skill by itself — a non-interactive, promptless install path an agent can run headlessly end to end
- Point an agent at llms.txt or agent-oriented docs
- Plug MCP servers into this product so it can use their tools
- Use an official CLI
- Delegate tasks to a built-in AI assistant inside the product
- Operate the product with natural-language commands
- Test against a sandbox environment without touching production data
- Define rules that trigger actions automatically on events
- Version, review, and roll back my automations
- Install the same collection into multiple different coding agents — Claude Code, Codex, Cursor, and others — with per-harness instructions
- Skills are plain markdown files and folders I can read, copy, and carry to another harness — not a proprietary binary format
- Browse or search a catalog of available skills — a registry, leaderboard, or marketplace listing — before installing anything
- Distribute a standard skill set to my whole team — via a marketplace, a shared repo, or files committed to the project
- A quickstart takes me from nothing to a working installed skill in under five minutes
- Every skill documents what it does and when it activates, so I can predict my agent's new behavior before it surprises me
- Install a skill collection with one documented command — a package-manager one-liner, CLI, or in-agent marketplace command — and it is active in my next session
- Choose install scope — project-local files committed with my repo, or user-global across all projects
- Install only the specific skills I want from a collection instead of taking the whole bundle
- Do everything through the API that I can do in the UI
- Review exactly what instructions and scripts a skill will add — list contents before installing and read every file afterward
- Skills follow the open Agent Skills specification so the same skill folder is valid beyond this one vendor's tooling
- Control when skill changes reach my team — pinned versions or a lockfile rather than silent behind-the-back updates
- There is a documented update path — marketplace auto-updates or an explicit update command — so I get fixes without reinstalling from scratch
GitHub README15 stories
- My agent can author and package a new skill end to end by following the project's own spec, template, or meta-skill
- My coding agent can install a skill by itself — a non-interactive, promptless install path an agent can run headlessly end to end
- Use an official CLI
- Get AI-generated insights and suggestions from my data inside the product
- Skills are plain markdown files and folders I can read, copy, and carry to another harness — not a proprietary binary format
- Browse or search a catalog of available skills — a registry, leaderboard, or marketplace listing — before installing anything
- Distribute a standard skill set to my whole team — via a marketplace, a shared repo, or files committed to the project
- Installed skills trigger automatically from task context, with descriptions engineered so the agent activates the right skill at the right moment
- A quickstart takes me from nothing to a working installed skill in under five minutes
- Install a skill collection with one documented command — a package-manager one-liner, CLI, or in-agent marketplace command — and it is active in my next session
- Install only the specific skills I want from a collection instead of taking the whole bundle
- Review exactly what instructions and scripts a skill will add — list contents before installing and read every file afterward
- Author a new skill from a documented template — a SKILL.md with name and description frontmatter — without reverse-engineering existing skills
- The collection ships a meta-skill or tool that guides my agent through writing, improving, and packaging new skills
- There is a documented update path — marketplace auto-updates or an explicit update command — so I get fixes without reinstalling from scratch
Specification docs11 stories
- My agent can author and package a new skill end to end by following the project's own spec, template, or meta-skill
- Download a machine-readable API spec (OpenAPI or equivalent)
- Perform bulk operations across many items at once
- Install the same collection into multiple different coding agents — Claude Code, Codex, Cursor, and others — with per-harness instructions
- Skills are plain markdown files and folders I can read, copy, and carry to another harness — not a proprietary binary format
- Installed skills trigger automatically from task context, with descriptions engineered so the agent activates the right skill at the right moment
- Every skill documents what it does and when it activates, so I can predict my agent's new behavior before it surprises me
- Review exactly what instructions and scripts a skill will add — list contents before installing and read every file afterward
- Author a new skill from a documented template — a SKILL.md with name and description frontmatter — without reverse-engineering existing skills
- The collection ships a meta-skill or tool that guides my agent through writing, improving, and packaging new skills
- Skills follow the open Agent Skills specification so the same skill folder is valid beyond this one vendor's tooling
Hacker News10 stories
- Plug MCP servers into this product so it can use their tools
- Get AI-generated insights and suggestions from my data inside the product
- Delegate tasks to a built-in AI assistant inside the product
- Operate the product with natural-language commands
- Perform bulk operations across many items at once
- Define rules that trigger actions automatically on events
- Installed skills trigger automatically from task context, with descriptions engineered so the agent activates the right skill at the right moment
- A quickstart takes me from nothing to a working installed skill in under five minutes
- Every skill documents what it does and when it activates, so I can predict my agent's new behavior before it surprises me
- Author a new skill from a documented template — a SKILL.md with name and description frontmatter — without reverse-engineering existing skills
docs9 stories
- My coding agent can install a skill by itself — a non-interactive, promptless install path an agent can run headlessly end to end
- Run the product headlessly / in CI for automation
- Drive the product through a documented public API
- Build against official SDKs
- Download a machine-readable API spec (OpenAPI or equivalent)
- Test against a sandbox environment without touching production data
- Version, review, and roll back my automations
- Do everything through the API that I can do in the UI
- Control when skill changes reach my team — pinned versions or a lockfile rather than silent behind-the-back updates
Engineering docs7 stories
- Get AI-generated insights and suggestions from my data inside the product
- Delegate tasks to a built-in AI assistant inside the product
- Operate the product with natural-language commands
- Perform bulk operations across many items at once
- Define rules that trigger actions automatically on events
- Installed skills trigger automatically from task context, with descriptions engineered so the agent activates the right skill at the right moment
- Review exactly what instructions and scripts a skill will add — list contents before installing and read every file afterward
Probe proofs — replayable recordings from the probe harnessProbe proofs
Replayable recordings from our probe harness — see the Prove-It protocol to submit one.
$npx -y skills add anthropics/skills --skill skill-creator -a claude-code -y --copy # in a scratch dir, then print installed SKILL.md frontmatterreproduced$ npx -y skills add anthropics/skills --skill skill-creator -a claude-code -y --copy # in a scratch dir, then print installed SKILL.md frontmatter \ ███████╗██╗ ██╗██╗██╗ ██╗ ███████╗ ██╔════╝██║ ██╔╝██║██║ ██║ ██╔════╝ ███████╗█████╔╝ ██║██║ ██║ ███████╗ ╚════██║██╔═██╗ ██║██║ ██║ ╚════██║ ███████║██║ ██╗██║███████╗███████╗███████║ ╚══════╝╚═╝ ╚═╝╚═╝╚══════╝╚══════╝╚══════╝ ┌ skills │ ◇ Source: https://github.com/anthropics/skills.git │ [ 3 5 m ◒ [ 3 9m C l o n i n g r e p o s i t o r y … [ 3 5 m ◐ [ 3 9m C l o n i n g r e p o s i t o r y … [ 3 5 m ◓ [ 3 9m C l o n i n g r e p o s i t o r y … [ 3 5 m ◑ [ 3 9m C l o n i n g r e p o s i t o r y … [ 3 5 m ◒ [ 3 9m C l o n i n g r e p o s i t o r y … [ 3 5 m ◐ [ 3 9m C l o n i n g r e p o s i t o r y … [ 3 5 m ◓ [ 3 9m C l o n i n g r e p o s i t o r y … [ 3 5 m ◑ [ 3 9m C l o n i n g r e p o s i t o r y … [ 3 5 m ◒ [ 3 9m C l o n i n g r e p o s i t o r y … . [ 3 5 m ◐ [ 3 9m C l o n i n g r e p o s i t o r y … . [ 3 5 m ◓ [ 3 9m C l o n i n g r e p o s i t o r y … . [ 3 5 m ◑ [ 3 9m C l o n i n g r e p o s i t o r y … . [ 3 5 m ◒ [ 3 9m C l o n i n g r e p o s i t o r y … . [ 3 5 m ◐ [ 3 9m C l o n i n g r e p o s i t o r y … . [ 3 5 m ◓ [ 3 9m C l o n i n g r e p o s i t o r y … . [ 3 5 m ◑ [ 3 9m C l o n i n g r e p o s i t o r y … . [ 3 5 m ◒ [ 3 9m C l o n i n g r e p o s i t o r y … . . [ 3 5 m ◐ [ 3 9m C l o n i n g r e p o s i t o r y … . . [ 3 5 m ◓ [ 3 9m C l o n i n g r e p o s i t o r y … . . [ 3 5 m ◑ [ 3 9m C l o n i n g r e p o s i t o r y … . . [ 3 5 m ◒ [ 3 9m C l o n i n g r e p o s i t o r y … . . [ 3 5 m ◐ [ 3 9m C l o n i n g r e p o s i t o r y … . . [ 3 5 m ◓ [ 3 9m C l o n i n g r e p o s i t o r y … . . [ 3 5 m ◑ [ 3 9m C l o n i n g r e p o s i t o r y … . . [ 3 5 m ◒ [ 3 9m C l o n i n g r e p o s i t o r y … . . . [ 3 5 m ◐ [ 3 9m C l o n i n g r e p o s i t o r y … . . . [ 3 5 m ◓ [ 3 9m C l o n i n g r e p o s i t o r y … . . . [ 3 5 m ◑ [ 3 9m C l o n i n g r e p o s i t o r y … . . . [ 3 5 m ◒ [ 3 9m C l o n i n g r e p o s i t o r y … . . .◇ Repository cloned │ ◇ Found 20 skills │ ● Selected 1 skill: skill-creator │ ◇ Installation Summary ─╮ │ │ │ │ │ │ │ │ │ [ │ │ 1 │ │ m │ │ E │ │ x │ │ a │ │ m │ │ p │ │ l │ │ e │ │ │ │ S │ │ k │ │ i │ │ l │ │ l │ │ s │ │ │ │ [ │ │ 2 │ │ 2m │ │ │ │ │ │ │ │ [ │ │ 3 │ │ 6 │ │ m │ │ . │ │ / │ │ . │ │ a │ │ g │ │ e │ │ n │ │ t │ │ s │ │ / │ │ s │ │ k │ │ i │ │ l │ │ l │ │ s │ │ / │ │ s │ │ k │ │ i │ │ l │ │ l │ │ - │ │ c │ │ r │ │ e │ │ a │ │ t │ │ o │ │ r │ │ │ │ [ │ │ 3 │ │ 9m │ │ │ │ │ │ │ │ │ │ [ │ │ 2 │ │ m │ │ c │ │ o │ │ p │ │ y │ │ │ │ → │ │ │ │ [ │ │ 2 │ │ 2m │ │ │ │ C │ │ l │ │ a │ │ u │ │ d │ │ e │ │ │ │ C │ │ o │ │ d │ │ e │ │ │ ├────────────────────────╯ │ ◇ Security Risk Assessments ─╮ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ [ │ │ 2 │ │ m │ │ G │ │ e │ │ n │ │ │ │ [ │ │ 2 │ │ 2m │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ [ │ │ 2 │ │ m │ │ S │ │ o │ │ c │ │ k │ │ e │ │ t │ │ │ │ [ │ │ 2 │ │ 2m │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ [ │ │ 2 │ │ m │ │ S │ │ n │ │ y │ │ k │ │ │ │ [ │ │ 2 │ │ 2m │ │ │ │ │ │ [ │ │ 3 │ │ 6 │ │ m │ │ s │ │ k │ │ i │ │ l │ │ l │ │ - │ │ c │ │ r │ │ e │ │ a │ │ t │ │ o │ │ r │ │ │ │ [ │ │ 3 │ │ 9m │ │ │ │ │ │ │ │ [ │ │ 3 │ │ 2 │ │ m │ │ S │ │ a │ │ f │ │ e │ │ │ │ [ │ │ 3 │ │ 9m │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ [ │ │ 3 │ │ 2 │ │ m │ │ 0 │ │ │ │ a │ │ l │ │ e │ │ r │ │ t │ │ s │ │ │ │ [ │ │ 3 │ │ 9m │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ [ │ │ 3 │ │ 2 │ │ m │ │ L │ │ o │ │ w │ │ │ │ R │ │ i │ │ s │ │ k │ │ │ │ [ │ │ 3 │ │ 9m │ │ │ │ │ │ │ │ [ │ │ 2 │ │ m │ │ D │ │ e │ │ t │ │ a │ │ i │ │ l │ │ s │ │ : │ │ │ │ [ │ │ 2 │ │ 2m │ │ │ │ │ │ [ │ │ 2 │ │ m │ │ h │ │ t │ │ t │ │ p │ │ s │ │ : │ │ / │ │ / │ │ s │ │ k │ │ i │ │ l │ │ l │ │ s │ │ . │ │ s │ │ h │ │ / │ │ a │ │ n │ │ t │ │ h │ │ r │ │ o │ │ p │ │ i │ │ c │ │ s │ │ / │ │ s │ │ k │ │ i │ │ l │ │ l │ │ s │ │ │ │ [ │ │ 2 │ │ 2m │ │ │ ├─────────────────────────────╯ │ ◇ Installation complete │ ◇ Installed 1 skill ─╮ │ │ │ │ │ │ │ │ │ [ │ │ 1 │ │ m │ │ E │ │ x │ │ a │ │ m │ │ p │ │ l │ │ e │ │ │ │ S │ │ k │ │ i │ │ l │ │ l │ │ s │ │ │ │ [ │ │ 2 │ │ 2m │ │ │ │ │ │ [ │ │ 3 │ │ 2 │ │ m │ │ ✓ │ │ │ │ [ │ │ 3 │ │ 9m │ │ │ │ s │ │ k │ │ i │ │ l │ │ l │ │ - │ │ c │ │ r │ │ e │ │ a │ │ t │ │ o │ │ r │ │ │ │ │ │ [ │ │ 2 │ │ m │ │ ( │ │ c │ │ o │ │ p │ │ i │ │ e │ │ d │ │ ) │ │ │ │ [ │ │ 2 │ │ 2m │ │ │ │ │ │ │ │ │ │ [ │ │ 2 │ │ m │ │ → │ │ │ │ [ │ │ 2 │ │ 2m │ │ │ │ . │ │ / │ │ . │ │ c │ │ l │ │ a │ │ u │ │ d │ │ e │ │ / │ │ s │ │ k │ │ i │ │ l │ │ l │ │ s │ │ / │ │ s │ │ k │ │ i │ │ l │ │ l │ │ - │ │ c │ │ r │ │ e │ │ a │ │ t │ │ o │ │ r │ │ │ ├─────────────────────╯ │ └ Done! Review skills before use; they run with full agent permissions. \--- installed SKILL.md frontmatter --- --- name: skill-creator description: Create new skills, modify and improve existing skills, and measure skill performance. Use when users want to create a skill from scratch, edit, or optimize an existing skill, run evals to test a skill, benchmark skill performance with variance analysis, or optimize a skill's description for better triggering accuracy.
Claims vs evidence — vendor claims reconciled against independent verdictsClaims vs evidence
2 of 14 testable claims verified · 4 contradicted → integrity 0/100
32 distinct capability claims found in Anthropic Skills’s own claimed-docs/GitHub materials, reconciled against our judge’s independent verdicts.
2
Verified
8
Unverified
4
Contradicted
18
Undersold
Verified (5)
“Author a skill by writing a SKILL.md with instructions; Claude automatically adds it to its toolkit”
Author a new skill from a documented template — a SKILL.md with name and description frontmatter — without reverse-engineering existing skillsfullproof ↗
“Skill frontmatter controls whether the user or Claude decides when the skill is invoked, and skills can include a directory of supporting files”
Author a new skill from a documented template — a SKILL.md with name and description frontmatter — without reverse-engineering existing skillsfullproof ↗
“The required description field (1-1024 chars) must state both what the skill does and when to use it”
Author a new skill from a documented template — a SKILL.md with name and description frontmatter — without reverse-engineering existing skillsfullproof ↗
“Skills are simple to create — just a folder containing a SKILL.md file with YAML frontmatter and instructions”
Author a new skill from a documented template — a SKILL.md with name and description frontmatter — without reverse-engineering existing skillsfullproof ↗
“Skills are simple to create — just a folder containing a SKILL.md file with YAML frontmatter and instructions”
A quickstart takes me from nothing to a working installed skill in under five minutespartialproof ↗
Unverified (16)
“Custom plugins can extend Claude Code by bundling skills, agents, hooks, and MCP servers together”
Distribute a standard skill set to my whole team — via a marketplace, a shared repo, or files committed to the projectfullproof ↗
“Plugins are self-contained directories invoked via /plugin-name:hello, enabling sharing with teammates, community distribution, and versioned releases”
Distribute a standard skill set to my whole team — via a marketplace, a shared repo, or files committed to the projectfullproof ↗
“A plugin marketplace acts as a catalog providing centralized discovery, version tracking, and automatic updates for plugins”
Browse or search a catalog of available skills — a registry, leaderboard, or marketplace listing — before installing anythingpartialproof ↗
“Plugins can be added and installed with simple commands: /plugin marketplace add and /plugin install”
Install a skill collection with one documented command — a package-manager one-liner, CLI, or in-agent marketplace command — and it is active in my next sessionpartialproof ↗
“Skills can be specified via the API's container parameter (skill_id, type, version) to run inside a code execution environment”
Drive the product through a documented public APIfullproof ↗
“A dedicated meta-skill exists for creating new skills and iteratively improving them”
The collection ships a meta-skill or tool that guides my agent through writing, improving, and packaging new skillsfullproof ↗
“Skills can bundle additional files within their directory and reference them by name from SKILL.md”
Skills are plain markdown files and folders I can read, copy, and carry to another harness — not a proprietary binary formatfullproof ↗
“Up to 20 skills can be included in a single Messages API request”
Drive the product through a documented public APIfullproof ↗
“Skills can be uploaded and managed through a dedicated Skills API”
Drive the product through a documented public APIfullproof ↗
“A skill is, at minimum, just a directory containing a SKILL.md file”
Skills are plain markdown files and folders I can read, copy, and carry to another harness — not a proprietary binary formatfullproof ↗
“A plugin manifest file (.claude-plugin/plugin.json) defines the plugin's name, description, and version”
Distribute a standard skill set to my whole team — via a marketplace, a shared repo, or files committed to the projectfullproof ↗
“Plugin marketplaces support multiple source types, including git repositories and local paths”
Browse or search a catalog of available skills — a registry, leaderboard, or marketplace listing — before installing anythingpartialproof ↗
“A new skill can be tested locally before wider distribution using the --plugin-dir flag”
Review exactly what instructions and scripts a skill will add — list contents before installing and read every file afterwardpartialproof ↗
“A GitHub repository of skills can be registered as a Claude Code plugin marketplace with a single command”
Install a skill collection with one documented command — a package-manager one-liner, CLI, or in-agent marketplace command — and it is active in my next sessionpartialproof ↗
“A GitHub repository of skills can be registered as a Claude Code plugin marketplace with a single command”
Browse or search a catalog of available skills — a registry, leaderboard, or marketplace listing — before installing anythingpartialproof ↗
“Specific skill collections (e.g. document-skills, example-skills) can be installed selectively rather than installing everything”
Install only the specific skills I want from a collection instead of taking the whole bundlepartialproof ↗
Contradicted (6)
“Claude can invoke a skill automatically when relevant, or a user can invoke one explicitly with /skill-name”
Installed skills trigger automatically from task context, with descriptions engineered so the agent activates the right skill at the right momentdisputedproof ↗
“Skill performance can be benchmarked with variance analysis, and descriptions optimized for better triggering accuracy”
The collection maintains tests or evals for its skills so changes are verified against regressions rather than shipped on vibesnoneproof ↗
“The required description field (1-1024 chars) must state both what the skill does and when to use it”
Every skill documents what it does and when it activates, so I can predict my agent's new behavior before it surprises medisputedproof ↗
“When Claude judges a skill relevant to the current task, it loads the skill by reading its full SKILL.md into context”
Installed skills trigger automatically from task context, with descriptions engineered so the agent activates the right skill at the right momentdisputedproof ↗
“Example skill descriptions specify both the capability (e.g. extracting/filling/merging PDFs) and the trigger conditions for when to use it”
Every skill documents what it does and when it activates, so I can predict my agent's new behavior before it surprises medisputedproof ↗
“Using the skills toolkit requires git, docker, jq, and internet access”
The project documents its security posture — what skills can execute, the trust model for third-party skills, and any telemetry or data collectionnone
Undersold (18)
My agent can author and package a new skill end to end by following the project's own spec, template, or meta-skillfullproof ↗
My coding agent can install a skill by itself — a non-interactive, promptless install path an agent can run headlessly end to endpartialproof ↗
Point an agent at llms.txt or agent-oriented docsfullproof ↗
Run the product headlessly / in CI for automationpartialproof ↗
Plug MCP servers into this product so it can use their toolspartialproof ↗
Get AI-generated insights and suggestions from my data inside the productpartialproof ↗
Download a machine-readable API spec (OpenAPI or equivalent)partialproof ↗
Test against a sandbox environment without touching production datapartialproof ↗
Perform bulk operations across many items at oncepartialproof ↗
Install the same collection into multiple different coding agents — Claude Code, Codex, Cursor, and others — with per-harness instructionspartialproof ↗
Choose install scope — project-local files committed with my repo, or user-global across all projectspartialproof ↗
Do everything through the API that I can do in the UIpartialproof ↗
Skills follow the open Agent Skills specification so the same skill folder is valid beyond this one vendor's toolingfullproof ↗
Control when skill changes reach my team — pinned versions or a lockfile rather than silent behind-the-back updatespartialproof ↗
There is a documented update path — marketplace auto-updates or an explicit update command — so I get fixes without reinstalling from scratchpartialproof ↗
Claims outside our story set (8)
Real capability claims found in Anthropic Skills’s own materials, but no story in this arena’s taxonomy covers them yet — that’s feedback on the taxonomy, not a mark against the product.
“A setting (disableBundledSkills) lets admins turn off all bundled skills except /doctor”
source ↗“/run and /verify commands work without setup, inferring how to launch a project from README, package.json, or Makefile”
source ↗“Built-in document skills are addressable by short IDs: pptx, xlsx, docx, pdf”
source ↗“Skills can be available to all users or kept private to a specific workspace”
source ↗“A bundled PDF skill gives Claude the ability to directly manipulate PDFs, such as filling out forms”
source ↗“Claude can build and run an app itself to verify a code change works, instead of relying only on tests or type checks”
source ↗“A command file at .claude/commands and a skill at .claude/skills both create the same slash command and behave identically”
source ↗“/run and /verify can be taught custom build and launch behavior specific to a project”
source ↗
Business model
Public skills repo: most example skills are Apache-2.0, the production document skills (docx/pdf/pptx/xlsx) are source-available; using skills inside Claude apps requires a paid Claude plan.
pricing ↗Score trend
How this product’s scores have moved as evidence and verdicts are re-derived — a point per change, not per day.
Try Experimental
Run it in the microterminal →Recorded agent sessions — and a live MCP handshake where the vendor ships one.
Flag
⚑ Flag a verdictThink a verdict is wrong? Opens a prefilled GitHub issue — or use the ⚑ next to any verdict above.
For agents
Agent surface uptime llms.txt 100% (30d, checked every 6h since Sep 8 '26)