agentleFS
MicrosoftKnown publishermicrosoft/cobol-docs-pipeline-sample/.github/skills/group-doc-finalizer/SKILL.md

group-doc-finalizer

Skill1 starsChanged 55 days agoMITView on GitHub

Author per-group `index.md` and `complete.md` for ONE group declared in `config/grouping.yaml`. LLM-driven.

The file

We split this skill into the rules it gives an agent

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

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈1

    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:

    All of the above are projected into the bundle by scripts/assemble-inputs.py so the LLM reads exactly one _doc-bundles/_finalize.bundle.md.

    line 44
  2. Distinctive:

    No file/line citations in prose.

    line 67

≈ “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

Status: active. Empty config/grouping.yaml (groups: []) is a graceful no-op for the whole phase — the stage and dispatch wrappers both exit 0.

Scope

This skill owns the phase-I per-group LLM authoring step. One invocation covers exactly bundle.languages in declared order at bundle.output_paths for a single group. Never add an unrequested English edition or intermediate. Structural parity applies only when multiple languages are requested.

Sibling skills it does NOT replace:

  • section-doc-finalizer — per-file index.md + complete.md (phase E).
  • requirements-group-writer — group ATE (phase J).
  • functional-analysis-group-writer — group AFU (phase K).

Inputs

  • config/grouping.yaml — group manifest (members, description).
  • docs/_shared/<member>/chunk-manifest.json — for each member.
  • docs/_shared/<member>/facts.json — TOC fields + incoming_xref[].
  • docs/<lang>/<member>/{index,complete}.md — the per-file deliverables produced by phase E in the first requested language (bundle.languages[0]). The stager requires these member inputs and embeds their excerpts as the source narrative for all requested output editions.

All of the above are projected into the bundle by scripts/assemble-inputs.py so the LLM reads exactly one _doc-bundles/_finalize.bundle.md.

Outputs

  • docs/_shared/_groups/<group_id>/_doc-bundles/_finalize.bundle.json
  • docs/_shared/_groups/<group_id>/_doc-bundles/_finalize.bundle.md
  • docs/<lang>/_groups/<group_id>/{index,complete}.md for each requested language only.

Procedure (planned)

  1. Stage the per-group finalize bundle from member artefacts.
  2. Dispatch the group-doc-finalizer agent with the bundle.
  3. The agent writes index.md (executive summary + cross-member TOC + incoming/outgoing call graph excerpts) and complete.md (stable anchored concatenation of member excerpts) per language.

Determinism rules

  • Member order in TOC and complete.md follows config/grouping.yaml.groups[*].members[*] order.
  • Anchors are <member-basename>__<section-id> (lowercase).
  • No file/line citations in prose; cite the member document instead.
  • complete.md MUST be a single self-contained file:
    • It MUST NOT link to external content. Every paragraph/section link MUST be an internal anchor (#<member-basename>__<section-id>) into the same file. Relative links to sibling docs (other *.md, *.mmd, _coverage.md, glossary, alt-language siblings) are delinked — keep the label text only.
    • It MUST NOT reference facts.json; use a generic "code" reference instead.
    • It MUST NOT contain an "Authoring self-check" paragraph or heading.

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.