agentleFS
UnfoldCRO-OrganizationUnfoldCRO-Organization/cro-agent-skills/skills/task-dispatch/SKILL.md

task-dispatch

Skill0 starsChanged 22 days agoMITView on GitHub

Decide whether a task requires direct work, one specialised capability, or high-complexity planning and independent reviews.

  • Commits and pushes

2 other projects use a version of this.

Adopted by

Most recently taken up by

The file

Show its rules and text

We split this skill into the rules it gives an agent

19 rules. About 0 are common, 16 are distinctive, and 3 are very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈16

    Distinctive · estimate

    This author's own choices. Worth a look.

  • 3

    Very specific

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

  • Common
  • Distinctive
  • Very specific
  1. Distinctive:

    If direct work is sufficient: use no extra skill, MCP, plugin or agent.

    line 34
  2. Very specific:

    If specialised capability is required: read only the relevant section of ~/.claude/capability-registry.md, read ~/.claude/agent.md only …

    line 35
  3. Distinctive:

    If complexity is below 80: proceed directly or with one required capability, perform relevant verification.

    line 36
  4. Distinctive:

    If complexity is 80+: state that work is high complexity, produce a short plan before major changes, use only necessary implementation …

    line 37
  5. Distinctive:

    When a task references another project, clone, migration, white-label build, imported content, fetched data, shared module or database …

    line 41
  6. Distinctive:

    Treat the source project as reference-only by default.

    line 44
  7. Distinctive:

    Do not copy data or use another project's database without explicit approval.

    line 45
  8. Distinctive:

    When a task involves git push, release, merge, version tag, deployment, rollback or production movement:

    line 51
  9. Distinctive:

    Treat it as sensitive and require explicit instruction before write actions.

    line 53
  10. Very specific:

    Read <project>/docs/16-DEPLOYMENT.md (+ its .local.md.

    line 54
  11. Distinctive:

    Do not invent branch, release, version, tag, server or deployment details.

    line 55
  12. Distinctive:

    Use only required git or deployment capability after approval.

    line 56
  13. Very specific:

    After actual release or deployment, add a release entry to docs/17-CHANGELOG.md and update docs/01-PRD.md §Scope status only where relevant …

    line 58
  14. Distinctive:

    Do not inspect the complete skill, plugin, MCP or agent inventory.

    line 62
  15. Distinctive:

    Do not invoke capability merely because it exists.

    line 63
  16. Distinctive:

    Do not call several similar capabilities to compare results by default.

    line 64
  17. Distinctive:

    Do not invoke deployment capability without explicit deployment instruction.

    line 65
  18. Distinctive:

    Do not invoke gstack or external packs unless the current required workflow justifies it.

    line 66
  19. Distinctive:

    Only when specialised capability or high-complexity process is required, report:

    line 70

≈ “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

Minimum Capability Task Dispatch

Requested task: $ARGUMENTS

Objective

Choose the smallest sufficient execution path without loading or invoking unrelated capabilities.

Do Not Use This Skill For

  • update md, update docs, save progress, update tasks or progress recording.
  • Simple project resolution handled by project-start.
  • New capability review handled by capability-intake.

Decision Process

  1. Identify the exact requested outcome.
  2. Identify the active project and affected work area.
  3. Determine whether direct work is sufficient.
  4. Assess task complexity internally from 0 to 100 using:
    • Number of modules affected.
    • Business-critical logic.
    • Authentication, permissions, finance, payments or sensitive-data impact.
    • External integrations.
    • Database or schema changes.
    • Production risk.
    • Unknowns, reversibility and regression risk.
  5. If direct work is sufficient: use no extra skill, MCP, plugin or agent.
  6. If specialised capability is required: read only the relevant section of ~/.claude/capability-registry.md, read ~/.claude/agent.md only when delegated review or execution is required, choose the smallest suitable capability, do not choose overlapping capabilities for the same purpose.
  7. If complexity is below 80: proceed directly or with one required capability, perform relevant verification.
  8. If complexity is 80+: state that work is high complexity, produce a short plan before major changes, use only necessary implementation capability, perform actual relevant verification, complete two independent review passes, resolve or report issues before declaring completion.

Cross-Project Data Check

When a task references another project, clone, migration, white-label build, imported content, fetched data, shared module or database reuse:

  1. Read ~/.claude/references/cross-project-data-copy.md.
  2. Treat the source project as reference-only by default.
  3. Do not copy data or use another project's database without explicit approval.
  4. Continue only with structure, logic or placeholder implementation where transfer is not approved.
  5. Increase risk assessment when private data, databases, credentials or production records are involved.

Release, Git Push and Deployment Check

When a task involves git push, release, merge, version tag, deployment, rollback or production movement:

  1. Treat it as sensitive and require explicit instruction before write actions.
  2. Read <project>/docs/16-DEPLOYMENT.md (+ its .local.md; unmigrated: docs/deploy.md) before proceeding.
  3. Do not invent branch, release, version, tag, server or deployment details.
  4. Use only required git or deployment capability after approval.
  5. For a major release, require an approved version tag before pushing or releasing.
  6. After actual release or deployment, add a release entry to docs/17-CHANGELOG.md and update docs/01-PRD.md §Scope status only where relevant (unmigrated: docs/changelog.md, docs/completed-vs-scope.md).

Capability Restriction

  • Do not inspect the complete skill, plugin, MCP or agent inventory.
  • Do not invoke capability merely because it exists.
  • Do not call several similar capabilities to compare results by default.
  • Do not invoke deployment capability without explicit deployment instruction.
  • Do not invoke gstack or external packs unless the current required workflow justifies it.

Dispatch Output

Only when specialised capability or high-complexity process is required, report:

Execution path: <Specialised capability / High-complexity workflow>
Complexity: <score only for substantial work or when 80+ rule applies>
Capability selected: <capability name only when selected>
Reason: <one concise line>

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.