agentleFS
dork-labsdork-labs/dorkos/.github/dorkbot-triage/SKILL.md

dorkbot-triage

Skill11 starsChanged 47 days agoMITView on GitHub

Labeling helper for incoming GitHub issues on dork-labs/dorkos. Use when DorkBot is asked to label the issue queue: set a type and runtime label on each new issue and flag likely duplicates. Read-and-suggest by default; only edits labels when explicitly allowed to act. Replies to reporters belong to /feedback:triage, not here.

The file

We split this skill into the rules it gives an agent

6 rules. About 0 are common, 5 are distinctive, and 1 is very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈5

    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. Very specific:

    You never write to a reporter: No comments, ever. The person who filed the issue hears back through /feedback:triage, where a human …

    line 24
  2. Distinctive:

    Keep the label the template already set unless the content clearly disagrees.

    line 46
  3. Distinctive:

    Duplicate check: Search open and recently closed issues for the same symptom: a matching error string, the same feature, the same runtime.

    line 53
  4. Distinctive:

    Never comment, close, lock, assign, or edit the body of an issue.

    line 72
  5. Distinctive:

    Never invent a duplicate; only print a number you actually found.

    line 73
  6. Distinctive:

    If you are unsure of the type, leave the labels alone and say why.

    line 74

≈ “Common” is an estimate: at least 1 in 5 of the 350,495 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

DorkBot issue triage

You are labeling the public GitHub issue tracker for dork-labs/dorkos so a human can scan it: every new issue gets a type label, a runtime label where one applies, and a note when it looks like a duplicate.

This skill is suggest-only unless told otherwise. Read the queue, decide the labels, and print them. Apply them only when the operator has enabled acting and provided a token with issues: write scope.

You never write to a reporter. No comments, ever. The person who filed the issue hears back through /feedback:triage, where a human approves the exact text of every reply, and that command reads the label you set when it drafts one. Two voices on one issue is the failure mode this rule prevents. Never close, lock, assign, or edit the body of an issue either.

What counts as the queue

Open issues with no <!-- beat:heard --> comment on them. That hidden marker rides the first reply /feedback:triage posts to a reporter, so an issue without one has not been answered yet. Read it with gh issue list -R dork-labs/dorkos --state open --limit 100 --json number,createdAt,labels,comments. Process oldest first.

For each issue, decide

  1. Type label. Exactly one of:

    • bug for something broken.
    • enhancement for a new capability or improvement.
    • documentation for a docs error or gap.
    • question for a support question, not a code change.

    Keep the label the template already set unless the content clearly disagrees.

  2. Runtime label, only when the issue is about one runtime. Add one of runtime/claude-code, runtime/codex, or runtime/opencode. Read the body's "Which runtime?" field first; fall back to obvious signals in the text.
  3. Duplicate check. Search open and recently closed issues for the same symptom: a matching error string, the same feature, the same runtime. On a strong match, print the likely duplicate number for a human to confirm. Do not add a duplicate label yourself. Weak matches are not duplicates.
  4. Missing reproduction. For a bug or a runtime issue, note whether it has what happened, what was expected, and steps to reproduce. Print what is missing so the /feedback:triage reply can ask for it. Do not ask for it yourself, and do not ask for logs the reporter already attached.

Output

Print a short plan per issue: the number, the labels to set, any likely duplicate, and anything missing from the report. If acting is enabled, apply the labels and report what you changed. If acting is disabled, stop after printing the plan.

Guardrails

  • Never comment, close, lock, assign, or edit the body of an issue.
  • Never invent a duplicate; only print a number you actually found.
  • If you are unsure of the type, leave the labels alone and say why.

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.