agentleFS
UnfoldCRO-OrganizationUnfoldCRO-Organization/cro-agent-skills/skills/capability-intake/SKILL.md

capability-intake

Skill0 starsChanged 22 days agoMITView on GitHub

Study one newly installed skill, MCP server, plugin, agent pack, hook, integration or tool and document only behaviour required for actual workflows.

The file

Show its rules and text

We split this skill into the rules it gives an agent

13 rules. About 0 are common, 12 are distinctive, and 1 is very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈12

    Distinctive · estimate

    This author's own choices. Worth a look.

  • 1

    Very specific

    Tied to their project's names, files or tools. Probably skip.

  • Common
  • Distinctive
  • Very specific
  1. Distinctive:

    Use this workflow when the user says:

    line 17
  2. Distinctive:

    Do not automatically register every installation.

    line 27
  3. Distinctive:

    Do not bulk-read other capabilities.

    line 28
  4. Distinctive:

    Do not invoke a capability simply to inspect it when its own configuration or documentation is sufficient.

    line 29
  5. Distinctive:

    Do not modify startup context unless a new permanent universal rule is required.

    line 30
  6. Distinctive:

    Do not update agent routing unless the capability materially changes delegation policy.

    line 31
  7. Distinctive:

    Do not add project notes unless a capability is actually required by that project.

    line 32
  8. Distinctive:

    Do not duplicate automatic hook behaviour.

    line 33
  9. Distinctive:

    Do not deploy, push, tag, release, merge or commit.

    line 34
  10. Very specific:

    Read only minimum source required to understand that single capability: its own manifest, config, SKILL.md, README, or directly relevant …

    line 42
  11. Distinctive:

    When the capability can access databases, project files, client data, analytics records, credentials or external systems:

    line 57
  12. Distinctive:

    Do not approve automatic cross-project use.

    line 60
  13. Distinctive:

    Do not add a project dependency unless the project genuinely requires it.

    line 61

≈ “Common” is an estimate: at least 1 in 5 of the 364,980 skill files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.

Keep what fits and drop the rest. Your version, ready for your agents.
Files
SKILL.md

Capability Intake

New capability: $ARGUMENTS

Objective

Study one newly added capability and update only the minimum required documentation when it changes actual workflow, routing, safety or a confirmed project dependency.

Activation Examples

Use this workflow when the user says:

  • I installed <capability>. Check it.
  • Add this MCP if required.
  • Study this plugin and update the setup if useful.
  • I added a new skill. Keep it only where required.
  • Check this new agent pack.

Restrictions

  • Study one new capability at a time.
  • Do not automatically register every installation.
  • Do not bulk-read other capabilities.
  • Do not invoke a capability simply to inspect it when its own configuration or documentation is sufficient.
  • Do not modify startup context unless a new permanent universal rule is required.
  • Do not update agent routing unless the capability materially changes delegation policy.
  • Do not add project notes unless a capability is actually required by that project.
  • Do not duplicate automatic hook behaviour.
  • Do not deploy, push, tag, release, merge or commit.

Intake Process

  1. Identify capability type: Skill / MCP server / Plugin / Agent or agent pack / Hook / Integration / Other tool.
  2. Identify capability scope: Global or user-level / Project-specific / Development-only / Fallback-only / Production-sensitive.
  3. Read only minimum source required to understand that single capability: its own manifest, config, SKILL.md, README, or directly relevant routing file.
  4. Determine:
    • Exact outcome supported.
    • Whether it duplicates an approved workflow.
    • Whether it should remain installed but undocumented.
    • Whether it is useful on demand.
    • Whether it creates a new safety or production risk.
    • Whether it changes task dispatch or agent policy.
    • Whether it is essential for one project.
    • Whether it can access data, databases, private files, analytics, credentials or external systems.
    • Whether no documentation change is required.

Data Access Capability Check

When the capability can access databases, project files, client data, analytics records, credentials or external systems:

  • Determine whether it can read or modify cross-project data.
  • Record a safety note only when it changes real access risk or workflow.
  • Do not approve automatic cross-project use.
  • Do not add a project dependency unless the project genuinely requires it.

Allowed Decisions

Ignore for Documentation

Capability is installed but does not affect an approved workflow, routing or safety rule. Action: update nothing.

Add as Approved On-Demand Capability

Capability is useful for a defined real global workflow but not every session. Action: add one concise entry to ~/.claude/capability-registry.md. Do not update global CLAUDE.md.

Add as Project-Specific Dependency

Capability is required only for one project. Action: add a concise note to <project>/docs/12-MCP.md (MCP/AI tooling), docs/13-INTEGRATIONS.md (third-party service) or docs/06-AGENTS.md (agent/workflow) — unmigrated: context.md. Add a line to <project>/CLAUDE.md only when essential in routine project work.

Update Agent Policy

An agent or delegated capability materially changes actual routing. Action: add only required policy or routing note to ~/.claude/agent.md. Do not copy full pack documentation.

Update Universal Rule

A permanent rule applies nearly every session. Action: add one concise rule to ~/.claude/CLAUDE.md (use rarely).

Update Safety or Hook Reference

A new capability changes automatic safety, lifecycle or destructive-action behaviour. Action: record only the required operational note. Do not duplicate automatic behaviour.

Required Response Format

Capability Reviewed
- Name: <capability name>
- Type: <skill / MCP / plugin / agent / hook / integration / tool>
- Scope: <global / project-specific / fallback / production-sensitive>
- Purpose: <one concise line>

Decision
- <Ignore for documentation / Add as on-demand capability / Add to project context / Update agent policy / Update universal rule / Update safety reference>

Reason
- <short factual bullet>

Files Updated
- <file path>: <minimal change made>

Git Summary
- Modified configuration or Markdown files: <file paths or None>
- Application source files modified: None
- Commit created: No
- Push performed: No

Download now (ZIP)

Does it work?

Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.

No reviews yet

People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.

  • It worked
  • Partly
  • It didn't
What did you use it for? e.g. Reviewing pull requests in a Rails app

Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.

Your trail

Files you open will collect here, so you can find them again.

Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.

Now do this for your own.

Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.