agentleFS
MicrosoftKnown publishermicrosoft/Employee-Self-Service-Agent-Developer-Kit/.github/agents/skills/triage-sample-issue/SKILL.md

triage-sample-issue

Skill29 starsChanged 6 months agoMITView on GitHub

When to use: any issue labeled area:samples, or whose body references samples/. Preconditions: you have read samples/AGENTS.md.

We split this skill into the rules it gives an agent

8 rules. About 0 are common, 6 are distinctive, and 2 are very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈6

    Distinctive · estimate

    This author's own choices. Worth a look.

  • 2

    Very specific

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

  • Common
  • Distinctive
  • Very specific
  1. Very specific:

    Read access to samples/.

    line 10
  2. Distinctive:

    New topic: must include:

    line 19
  3. Distinctive:

    Fix topic: must include:

    line 25
  4. Distinctive:

    Duplicate check: Before classifying, search open issues for overlap:

    line 29
  5. Distinctive:

    Treat as a duplicate when another open issue targets the same path (fix) or the same area + topic intent (new).

    line 33
  6. Distinctive:

    Stop and ask if the area, sub-grouping, or topic folder is ambiguous.

    line 58
  7. Very specific:

    Stop if the request would touch anything outside samples/.

    line 59
  8. Distinctive:

    Stop if the request asks to rename or delete existing files/folders.

    line 60

≈ “Common” is an estimate: at least 1 in 5 of the 308,232 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

Skill: Triage a samples GitHub issue

When to use: any issue labeled area:samples, or whose body references samples/.

Preconditions: you have read samples/AGENTS.md.

Inputs

  • Issue title, body, labels, author, linked items.
  • Read access to samples/.

Procedure

  1. Determine request kind.
    • new-topic — issue asks for a topic that does not exist under samples/.
    • fix-topic — issue references an existing path under samples/.
  2. Identify target. Area (ISV), sub-grouping (Employee / Manager / Extended / flat), topic folder name. Verify the path actually exists for fix-topic.
  3. Required-info check.
    • New topic must include:
      • Target area / ISV.
      • Proposed topic folder name (or enough context to derive one).
      • User scenario / expected behavior.
      • At least 2 sample user utterances.
      • Required inputs/outputs or referenced API/system of record.
    • Fix topic must include:
      • Target path under samples/ (folder or file).
      • Current behavior (what is wrong).
      • Expected behavior or example.
  4. Duplicate check. Before classifying, search open issues for overlap:
    • Query open issues with label area:samples plus keywords from the title and any samples/ path referenced in the body.
    • For fix-topic, also search by the exact target path under samples/.
    • For new-topic, also search by area/ISV + proposed topic folder name.
    • Treat as a duplicate when another open issue targets the same path (fix) or the same area + topic intent (new). Record the matching issue number.
  5. Classify into exactly one of:
    • actionable-new-topic
    • actionable-fix
    • needs-clarification — required info missing.
    • blocked — depends on something outside repo control.
    • duplicate — another open issue covers it.
    • out-of-scope — not about samples/, or asks for renames, deletions, cross-area refactors, or changes outside samples/.
    • not-actionable — spam/empty.
  6. Act on the classification:
    • actionable-new-topic / actionable-fix → apply agent:eligible, hand off to create-or-update-sample-topic.
    • needs-clarification → apply agent:needs-info, post a comment listing the specific missing fields. Do not modify files.
    • blocked → apply agent:blocked, post a short comment naming the external dependency and referencing samples/AGENTS.md. Do not modify files.
    • out-of-scope → apply agent:out-of-scope, post a short comment referencing the relevant rule in samples/AGENTS.md (safety boundaries or area scope). Do not modify files.
    • duplicate → apply the stock duplicate label, post a short comment linking the original issue (Duplicate of #<n>). Do not modify files. (No dedicated agent:duplicate label is defined.)
    • not-actionable → do not apply any label. Stop without commenting; optionally flag for human review. Do not modify files.

Outputs

  • Classification label.
  • Target path (for actionable cases).
  • Either: handoff packet to the next skill, OR a posted comment, OR a labeled stop.

Stop conditions

  • Stop and ask if the area, sub-grouping, or topic folder is ambiguous.
  • Stop if the request would touch anything outside samples/.
  • Stop if the request asks to rename or delete existing files/folders.

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.

Reviews can't be read right now.

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.