agentleFS
thefrederiksenthefrederiksen/cc-director/.claude/skills/document-features/skill.md

document-features

Skill96 starsChanged 16 days agoMITView on GitHub

Generate and refresh the public feature documentation in docs/public/features/ - scan the code to decide what is implemented, screenshot the running app, and update the pages. Triggers on "/document-features", "document the features", "refresh feature screenshots", "regenerate feature docs".

We split this skill into the rules it gives an agent

9 rules. About 0 are common, 2 are distinctive, and 7 are very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈2

    Distinctive · estimate

    This author's own choices. Worth a look.

  • 7

    Very specific

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

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

    Keep docs/public/features/ and its screenshots in sync with the app.

    line 8
  2. Very specific:

    Keep every source: path repo-relative and real.

    line 35
  3. Distinctive:

    Run it from an INTERACTIVE desktop session.

    line 53
  4. Very specific:

    Default dummy sessions are plain shells (-Agent RawCli) - no Claude subscription spent.

    line 57
  5. Very specific:

    Read the harness summary. It prints which inventory screens were NOT auto-captured.

    line 60
  6. Very specific:

    Follow the writing style of the existing docs/public/ pages: one-sentence intro, then the detail.

    line 75
  7. Very specific:

    Update docs/public/index.json so every page under docs/public/features/ has an entry in the features category.

    line 80
  8. Very specific:

    Only touches docs/features/, docs/public/features/, docs/public/index.json, and the capture script.

    line 99
  9. Distinctive:

    If a required tool (cc-click, git, dotnet) is missing, the harness fails loudly with the fix - do not work around it.

    line 102

≈ “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

Document Features

Keep docs/public/features/ and its screenshots in sync with the app. This is the "docs-as-code with regenerated screenshots" pipeline: the inventory is decided from source, the screenshots are captured from a real (throwaway) build, and the pages are regenerated together. Re-run after UI changes - never hand-edit screenshots.

Inputs and outputs

ArtifactRole
docs/features/feature-inventory.yamlSource-of-truth: each feature, its status, and the file that proves it exists
scripts/capture-feature-screenshots.ps1Launches a slot build, makes dummy sessions, screenshots each screen
docs/public/features/*.mdThe published pages (raw markdown on GitHub)
docs/public/features/assets/*.pngCommitted screenshots referenced by the pages
docs/public/index.jsonManifest listing every page

Workflow

STEP 1: Inventory pass (decide what is implemented)

  1. Read docs/features/feature-inventory.yaml.
  2. Scan the UI source the way /cencon-generate reasons over the solution. Look in src/CcDirector.Avalonia/ (views, Controls/, *Dialog.axaml) and phone/CcDirectorClient/ (phone pages).
  3. Reconcile:
    • A user-facing view/dialog that exists in source but is NOT in the inventory -> add it.
    • An inventory entry whose source file no longer exists -> remove it (or mark planned).
    • A surface that is stubbed (TODO / NotImplemented / framework-only) -> partial.
  4. Keep every source: path repo-relative and real. Keep ids stable (they are the screenshot stems). ASCII only.
  5. Report what changed (new / removed / status-changed features) before continuing.

STEP 2: Capture pass (screenshot the running app)

Run the harness:

powershell -NoProfile -ExecutionPolicy Bypass -File scripts/capture-feature-screenshots.ps1
  • It builds slot 5, launches via the cc-director-launch scheduled task (per CLAUDE.md - never spawn the app from this process tree for session-creating runs), creates dummy sessions, captures the shots, and tears down ONLY slot 5.
  • Screens are captured with cc-click screenshot --pid <slot pid>, which uses Win32 PrintWindow to grab OUR window's own pixels - occlusion-proof, so it can never accidentally capture one of the user's other CC Director windows.
  • Run it from an INTERACTIVE desktop session. The main-window and tab shots work regardless, but the modal dialog shots (New Session, Settings) only render when the app can be foregrounded - a background/non-interactive run reports them as missing rather than capturing them.
  • Default dummy sessions are plain shells (-Agent RawCli) - no Claude subscription spent. Pass -Agent ClaudeCode when you need real agent output for the wingman/voice screens.
  • Read the harness summary. It prints which inventory screens were NOT auto-captured; those need either a new row in the script's $Shots table or a manual capture.

To extend automated coverage, add a row to $Shots in the harness (each row is a screenshot name plus the cc-click navigation to reach that screen).

STEP 3: Document pass (regenerate the pages)

For each category in the inventory, update its page under docs/public/features/:

  • Embed each captured screenshot with a relative path: ![Alt](assets/<name>.png).
  • For a feature whose screenshot does not exist yet, write the text and add a line _Screenshot pending - <how to capture it>._ instead of a broken image link.
  • Regenerate the feature matrix table in 01-overview.md from the inventory (Feature | Status | Page).
  • Follow the writing style of the existing docs/public/ pages: one-sentence intro, then the detail. No filler, no Unicode.

STEP 4: Manifest pass

Update docs/public/index.json so every page under docs/public/features/ has an entry in the features category. Validate (same rules as /update-docs): every file points at a real file, every page is listed, ids are lowercase-kebab-case.

STEP 5: Report

Summarize: features added/removed/changed status, screenshots refreshed, screens still pending. Do NOT commit unless the user explicitly asks (repo rule).

Maintenance model

  • Re-run this skill before a release or whenever the README Features change.
  • The CI workflow .github/workflows/docs-drift.yml runs only STEP 1's reconcile (no GUI) on pull requests and flags when the inventory is stale relative to source. It cannot screenshot a desktop GUI on a headless runner - it only nudges a human to re-run the full local pass.

Notes

  • Only touches docs/features/, docs/public/features/, docs/public/index.json, and the capture script. Never product code under src/.
  • The desktop window title cc-click matches is "CC Director".
  • If a required tool (cc-click, git, dotnet) is missing, the harness fails loudly with the fix - do not work around it.

Skill Version: 1.0 Last Updated: 2026-06-19

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.