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

chronicle

Skill1 starsChanged 52 days agoApache-2.0View on GitHub

Have the AI draft or update a standalone Markdown doc — a feature write-up, design doc, or a section of README/ROADMAP — grounded in the real diff, code, commit history, or PR/issue text. Use when the user says chronicle, asks to write documentation, draft a design doc, or update a named doc file such as README or ROADMAP.

8 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

18 rules. 0 are common, 12 are distinctive, and 6 are very specific to this project.

  • 0

    Common

    Most of the 8 edited versions we read keep these.

  • 12

    Distinctive

    This author's own choices. Worth a look.

  • 6

    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 asks the AI to write or update a standalone Markdown document — anything that isn't an in-code comment or …

    line 8
  2. Very specific:

    Read context/policies/docs-policy.md for the shape rules.

    line 27
  3. Distinctive:

    Not permission to invent rationale, history, or intent the repo doesn't actually show.

    line 46
  4. Distinctive:

    Run when the user explicitly asks to chronicle, write documentation, draft a design doc or feature write-up, or update a doc file (README, …

    line 53
  5. Distinctive:

    If both the target doc and the thing it should document are ambiguous, ask which before writing anything.

    line 56
  6. Very specific:

    Read context/policies/docs-policy.md.

    line 71
  7. Distinctive:

    Check git status --short and the relevant diff to know what actually changed before drafting anything grounded in it.

    line 74
  8. Distinctive:

    Read the target document's existing structure and tone (or the closest comparable doc in the repo, when creating a new file) before writing …

    line 76
  9. Very specific:

    Read context/personality.md (target repo copy, else bundled) to confirm whether the target file/section is one where the orc voice actually …

    line 79
  10. Distinctive:

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

    line 112
  11. Very specific:

    Never touch in-code comments or docstrings — that's lore's job, not this one's.

    line 113
  12. Distinctive:

    Never invent rationale, history, features, or claims not grounded in a real, readable source.

    line 115
  13. Distinctive:

    Don't pad a doc with generic filler that carries no real information.

    line 117
  14. Distinctive:

    Stop and ask if there's no identifiable target and no current work to infer one from.

    line 122
  15. Distinctive:

    Stop only when the requested content depends entirely on a "why" that isn't recoverable from any real source, leaving nothing factual to …

    line 124
  16. Very specific:

    When only part of the content needs that rationale, don't stop — document the observable behavior that is grounded, and mark the specific …

    line 126
  17. Very specific:

    Stop if the user asked for a PR/issue-grounded target and neither the GitHub MCP connector nor gh is available — ask whether to use the …

    line 130
  18. Distinctive:

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

    line 145

“Common” is measured: kept by at least 6 in 10 of the 8 edited versions of this file we could read. “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

Chronicle Workflow

Use this skill when the user asks the AI to write or update a standalone Markdown document — anything that isn't an in-code comment or docstring.

Contents

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

At A Glance

  1. Read context/policies/docs-policy.md for the shape rules.
  2. Figure out the target: a diff/PR/feature to document, an existing doc file to update, or a new one to create.
  3. Ground every claim in something readable — code, commit history, or PR/issue text. Never invent intent, history, or rationale that isn't recoverable from a real source.
  4. Draft or update the doc content directly in the file.
  5. Report what changed, file and section. Never commit, push, or open/update a PR — that's ship's job.

What This Is Not

  • Not ship: it never stages, commits, or pushes. Run ship afterward to land the doc changes.
  • Not lore: lore writes in-code comments and docstrings; chronicle writes standalone Markdown documents — README, ROADMAP, design docs, feature write-ups — and never touches a source file's comments.
  • Not yap: yap explains in chat, read-only, and produces no file. This skill produces or updates an actual document.
  • Not permission to invent rationale, history, or intent the repo doesn't actually show. If the "why" behind a change isn't recoverable from code, commit history, or PR/issue text, say so and ask instead of narrating a plausible-sounding reason.

Trigger Rules

  • Run when the user explicitly asks to chronicle, write documentation, draft a design doc or feature write-up, or update a doc file (README, ROADMAP, or any other standalone Markdown doc).
  • If both the target doc and the thing it should document are ambiguous, ask which before writing anything.

Requirements

  • A diff, file, or new-file target needs nothing beyond local git and file access.
  • A PR or issue target additionally needs a GitHub MCP connector configured for this session (read access is enough — this skill never posts to GitHub), or gh installed and authenticated as a fallback. Without either, stop and tell the user the PR/issue target isn't reachable and ask whether to use the local diff or a named file instead.

Prerequisites

  • Read context/policies/docs-policy.md. Use the target repo's copy when it exists; otherwise use the bundled copy at ${CLAUDE_PLUGIN_ROOT}/context/policies/docs-policy.md.
  • Check git status --short and the relevant diff to know what actually changed before drafting anything grounded in it.
  • Read the target document's existing structure and tone (or the closest comparable doc in the repo, when creating a new file) before writing — match it instead of imposing a different convention.
  • Read context/personality.md (target repo copy, else bundled) to confirm whether the target file/section is one where the orc voice actually applies — most document content stays plain and technical; the voice is reserved for the specific surfaces that file lists (e.g. a README introduction), not for design docs or ROADMAP entries.

Targets

  • Diff, PR, or named feature/change — ground the doc content in what actually changed: git diff/git diff --staged plus untracked files, or a PR's diff fetched through the GitHub MCP connector (falling back to gh).
  • Existing file or section — update a named doc file or one section of it, matching its current structure and tone.
  • New file — create a doc that doesn't exist yet (e.g. a first CONTRIBUTING.md), grounded in the same real sources as any other target. Never fill a new doc with placeholder filler ("add details here") where a real source doesn't yet cover something — say what's missing instead.

Default Path

  1. Determine the target and read context/policies/docs-policy.md.
  2. Identify what the doc needs to say, grounded in the diff, code, commit history, or PR/issue text — not assumption or house style guessing.
  3. Match the existing document's structure and tone, or the plain, technical default from context/policies/docs-policy.md when creating a new file with no comparable precedent in the repo.
  4. Draft or update the content directly in the file.
  5. Report the changes as a file/section list.

Guardrails

  • Never commit, push, stage, or touch a PR/issue as part of this skill.
  • Never touch in-code comments or docstrings — that's lore's job, not this one's.
  • Never invent rationale, history, features, or claims not grounded in a real, readable source.
  • Don't pad a doc with generic filler that carries no real information; every sentence should say something specific about this repo.

Stop Conditions

  • Stop and ask if there's no identifiable target and no current work to infer one from.
  • Stop only when the requested content depends entirely on a "why" that isn't recoverable from any real source, leaving nothing factual to write. When only part of the content needs that rationale, don't stop — document the observable behavior that is grounded, and mark the specific gap inline (per context/policies/docs-policy.md) instead of guessing or blocking the rest.
  • Stop if the user asked for a PR/issue-grounded target and neither the GitHub MCP connector nor gh is available — ask whether to use the diff or a named file instead of silently switching targets.

Handoff

  • Unless the user asked for plain output, open with a fresh one-line phrase in the orc voice, speaking as the Chronicler 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 result instead. This orc-voice line is the skill's own success message, not part of the drafted document's content.
  • List the files/sections touched and a one-line summary of what changed.
  • Remind the user that ship is the next step if they want these changes committed.
  • 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.

Doc drafted:

📚 Hrrm. The feature's story is bound into the chronicle now. ✨

`docs/offline-cache.md`: new design doc, grounded in PR #58's diff and its
linked issue.
Run `ship` to carry this into a commit.

Doc updated:

📚 The scrolls now read true. README's Install section matches what's
actually shipped.

Blocked, no target:

❓ Chief, what should the chronicle cover, and where does it belong —
README, ROADMAP, or a new doc?

Doc drafted with a gap marked inline, not blocked:

📚 Hrrm. The retry logic's bound into the chronicle, with one gap marked. ✨

`docs/retry-policy.md`: documented the retry behavior; the 3-attempt limit
is left as an open question since nothing in the diff, history, or the
linked issue explains why 3 specifically.

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.