agentleFS
karloowskarloows/orchraft/skills/runes/SKILL.md

runes

Skill1 starsChanged 52 days agoApache-2.0View on GitHub

Have the AI summarize where the current branch's work actually stands — its PR, checks, and unresolved review findings — grounded in live git/GitHub state, not memory. Use when the user says runes, asks to resume, "where were we", "what's the status", or wants to pick a branch back up without re-explaining it.

1 other project uses a version of this.

Adopted by

Most recently taken up by

  • Other versions1 project uses a version of this.

    Word for word · 0

    None found.

    Edited · 0

    None found.

    Wrote their own version · 1

    1. aibtcdev/skills★ 11same name

    How matches are found. Word for word: the exact same text. Edited: shares passages with this file (“99% same” means 99% of its five-word phrases also appear here; 10% or more counts). Their own version: similar meaning, or, weakest, the same name. We can’t tell which project had it first. Projects are counted among the 200 most-starred candidates, and mentions are links to the repository we found, up to 50 per site, so both counts are a floor.

  • In this repository19 other files it gives its agents.

The file

Show its rules and text

We split this skill into the rules it gives an agent

25 rules. About 0 are common, 12 are distinctive, and 13 are 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.

  • 13

    Very specific

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

  • Common
  • Distinctive
  • Very specific
  1. Distinctive:

    Use this skill when the user wants to pick up existing work without re-explaining it, and wants the status grounded in the actual current …

    line 8
  2. Very specific:

    Read its live state fresh: local worktree, its pull request (if any), check/mergeability status, and the latest roast review's standing …

    line 30
  3. Distinctive:

    Stop. Never commit, push, or touch a PR/issue as part of this skill.

    line 35
  4. Very specific:

    Not an orchestrator: naming the next logical step is not running it.

    line 43
  5. Distinctive:

    Not a substitute for checking approval-policy.md before a mutation — this skill never mutates, so it carries no approval of its own.

    line 46
  6. Distinctive:

    Run when the user explicitly asks to runes, resume, "where were we"/ "what's the status"/"catch me up", or wants to pick a branch back up …

    line 51
  7. Distinctive:

    If no branch is current and none is named, ask which branch to summarize instead of guessing.

    line 54
  8. Very specific:

    Do not run automatically at the start of every session.

    line 56
  9. Very specific:

    Look up whether the target branch has a pull request via the GitHub MCP connector (falling back to gh when unavailable or blocked), by …

    line 65
  10. Distinctive:

    If a PR exists in any state, fetch its current state fresh in this turn: head SHA, mergeable state, check runs/status, and its most recent …

    line 72
  11. Very specific:

    If the target branch is the currently checked-out branch, check git status --short for local uncommitted changes, since those affect what …

    line 82
  12. Very specific:

    If the PR state is none, report that plainly — the next step is ship, not a status summary of a PR that doesn't exist.

    line 92
  13. Very specific:

    If the PR state is draft, report it as a draft explicitly — the next step is finishing and marking it ready, not ship (which would try to …

    line 94
  14. Very specific:

    If the PR state is closed or merged, report that plainly — the branch is stale relative to it.

    line 97
  15. Very specific:

    If the PR state is open, gather: PR number/URL, check/mergeability status, the latest roast review's finding counts by severity, and …

    line 100
  16. Distinctive:

    Keep it to what's actually known — don't pad a thin status with filler.

    line 113
  17. Distinctive:

    Never commit, push, or touch a PR/issue as part of this skill.

    line 127
  18. Very specific:

    Never present a roast review's findings as current once the PR head has moved past that review's marker commit — say they may be stale …

    line 128
  19. Distinctive:

    Never invent a status for something that couldn't be checked (e.g.

    line 130
  20. Distinctive:

    Never attribute the currently checked-out worktree's uncommitted changes to a target branch that isn't actually checked out.

    line 132
  21. Very specific:

    Never recommend ship as the next step without a confirmed none PR state — a missed PR lookup is not the same as no PR existing.

    line 134
  22. Distinctive:

    Stop and ask which branch to summarize if there's no current branch and none was named.

    line 139
  23. Very specific:

    Stop if the GitHub MCP connector and gh are both unavailable and the branch has a PR to check — report the local-only state (branch, …

    line 141
  24. Very specific:

    Stop after presenting the summary.

    line 144
  25. Distinctive:

    If blocked, drop the orc voice and state the blocker plainly.

    line 155

≈ “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

Runes Workflow

Use this skill when the user wants to pick up existing work without re-explaining it, and wants the status grounded in the actual current state of the branch and its pull request, not a recollection from earlier in the conversation.

Contents

  • At A Glance
  • What This Is Not
  • Trigger Rules
  • Prerequisites
  • Default Path
  • Summary Format
  • Guardrails
  • Stop Conditions
  • Handoff
  • Response Examples

At A Glance

  1. Identify the branch to report on (current branch by default, or one the user names).
  2. Read its live state fresh: local worktree, its pull request (if any), check/mergeability status, and the latest roast review's standing findings.
  3. Summarize where it actually stands and name the next logical step, without taking it.
  4. Stop. Never commit, push, or touch a PR/issue as part of this skill.

What This Is Not

  • Not yap: yap explains a specific named thing in depth; runes gives a status snapshot of a branch's whole lifecycle position.
  • Not roast: it doesn't produce new findings — it reports the standing result of the last roast pass already on the PR.
  • Not an orchestrator: naming the next logical step is not running it. Any actual next step (ship, roast, land) still needs the user to ask for it in that turn, per context/policies/approval-policy.md.
  • Not a substitute for checking approval-policy.md before a mutation — this skill never mutates, so it carries no approval of its own.

Trigger Rules

  • Run when the user explicitly asks to runes, resume, "where were we"/ "what's the status"/"catch me up", or wants to pick a branch back up without restating its history.
  • If no branch is current and none is named, ask which branch to summarize instead of guessing.
  • Do not run automatically at the start of every session; only when the user asks, or a project-local hook explicitly invokes it (see context/policies/approval-policy.md — a hook surfacing status is not the same as it mutating anything).

Prerequisites

  • Determine the target branch: the current branch by default, or the one the user names.
  • Look up whether the target branch has a pull request via the GitHub MCP connector (falling back to gh when unavailable or blocked), by branch name — not by whether the local branch has an upstream configured, since a PR can exist on GitHub even when the local branch has no upstream tracking ref set. Classify the result explicitly: none, draft, open, closed, or merged. Only a confirmed none justifies naming ship as the next step.
  • If a PR exists in any state, fetch its current state fresh in this turn: head SHA, mergeable state, check runs/status, and its most recent reviews. Search both PR reviews and issue comments for roast's delta-tracking marker (<!-- roast:review head=<sha> -->) from its Comment Format — roast falls back to posting as an issue comment when its review-write path is unavailable, so a fallback report would carry the marker there instead of in a review. If both a review and an issue comment carry the marker, use the one posted most recently. Use its marker to find the latest roast pass's standing findings and whether the PR head has moved past it.
  • If the target branch is the currently checked-out branch, check git status --short for local uncommitted changes, since those affect what ship would actually do next. If the target branch isn't currently checked out, its own uncommitted-changes state can't be read without switching to it (this skill doesn't check out branches) — say that plainly instead of attributing the current worktree's changes to it.

Default Path

  1. Resolve the target branch (see Prerequisites).
  2. If the PR state is none, report that plainly — the next step is ship, not a status summary of a PR that doesn't exist.
  3. If the PR state is draft, report it as a draft explicitly — the next step is finishing and marking it ready, not ship (which would try to open a duplicate) or land.
  4. If the PR state is closed or merged, report that plainly — the branch is stale relative to it; the next step is likely starting fresh or cleaning up the branch, not shipping more commits onto it.
  5. If the PR state is open, gather: PR number/URL, check/mergeability status, the latest roast review's finding counts by severity, and whether any of its threads are still unresolved.
  6. Note whether the PR head has moved since that roast review's marker commit — if so, say the standing findings may be stale rather than presenting them as current.
  7. Note local uncommitted changes per the Prerequisites' checked-out-branch caveat.
  8. Name the next logical step (e.g. "roast hasn't run yet", "2 findings still unresolved", "clean and land-ready") without taking it.

Summary Format

Keep it to what's actually known — don't pad a thin status with filler.

  • Branch — name and whether it has an upstream.
  • PR — number, URL, state (none/draft/open/closed/merged), mergeable/check status.
  • Reviews — the latest roast pass's finding counts by severity, and whether that pass is still current for the PR's head commit. Omit unless the PR state is open.
  • Local changes — any uncommitted work, only when the target branch is the currently checked-out branch (see Prerequisites).
  • Next step — the logical next action, named but not taken.

Guardrails

  • Never commit, push, or touch a PR/issue as part of this skill.
  • Never present a roast review's findings as current once the PR head has moved past that review's marker commit — say they may be stale instead.
  • Never invent a status for something that couldn't be checked (e.g. no GitHub access) — say what's unverifiable instead of guessing.
  • Never attribute the currently checked-out worktree's uncommitted changes to a target branch that isn't actually checked out.
  • Never recommend ship as the next step without a confirmed none PR state — a missed PR lookup is not the same as no PR existing.

Stop Conditions

  • Stop and ask which branch to summarize if there's no current branch and none was named.
  • Stop if the GitHub MCP connector and gh are both unavailable and the branch has a PR to check — report the local-only state (branch, worktree) and say PR/review status couldn't be verified.
  • Stop after presenting the summary; do not proceed into ship, roast, or land without the user separately asking for it.

Handoff

  • Unless the user asked for plain output, open with a fresh one-line phrase in the orc voice, speaking as the Rune-Reader from context/personality.md (the target repo's copy when it exists, otherwise ${CLAUDE_PLUGIN_ROOT}/context/personality.md). If they did, open with the plain summary instead.
  • Present the summary per Summary Format.
  • If blocked, drop the orc voice and state the blocker plainly.

Response Examples

Opening lines are samples of the orc voice; write a fresh one each time.

PR with open findings:

📖 Hrrm. The runes say PR #142 still carries one open crack. Picking up there. ✨

Branch: `fix/login-crash` (PR #142, open)
PR: mergeable, checks green
Reviews: last roast pass (current) found 1 important, 0 critical
Next step: fix the finding, then ship and resolve its thread.

Clean and land-ready:

📖 Hah! The board reads clean. PR #145 stands ready for the gate.

Branch: `docs/land-workflow` (PR #145, open)
PR: mergeable, checks green
Reviews: last roast pass (current) found 0 issues
Next step: land whenever you give the word.

No PR yet:

📖 Hrrm. This branch hasn't taken the field yet.

Branch: `feat/offline-cache` (no PR)
Local changes: 3 files modified, not yet committed
Next step: ship it to open the PR.

Blocked, no branch to check:

❓ Chief, which branch should I read the runes on?

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.