Skip to content

v1.3.0

Choose a tag to compare

View all tags
@Freakling Freakling released this 09 Oct 12:32
· 55 commits to main since this release
v1.3.0
3bd7c23

Designed to work with any AI coding assistant that reads AGENTS.md. Built and tested on Claude Code; other assistants are untested.

How to install

Your app needs git (git init if it has none), bash (on Windows: Git for Windows), and gitleaks >= 8.18.0 (macOS: brew install gitleaks; Windows: winget install gitleaks).

Option 1: npx skill

npx skills add Freakling/App-Director

Ask your assistant to set up App Director, or in Claude Code run /app-director.

Option 2: Claude Code plugin

/plugin marketplace add Freakling/App-Director
/plugin install app-director@app-director

Open a new session in your app's folder (or run /reload-plugins), then run /app-director:app-director.

Option 3: manual clone (next to your app as App-Director/)

git clone https://github.com/Freakling/App-Director.git App-Director

Then ask your assistant: Read App-Director/ONBOARDING.md and follow it.

Upgrading: Run /app-director (or ask your assistant to run onboarding). It detects an existing install and applies only what has changed. From 1.2.x no manual steps are needed; restart Claude Code afterwards so /drift-reset loads. From 1.1.x, also follow the 1.2.0 upgrade steps (install gitleaks, run bash tools/setup-clone.sh in every clone).

What's in 1.3.0

/drift-reset: the design drift reset. When a principle in your brief has been built more than one way, patching one place makes the other worse. The reset rebuilds shared understanding instead. Align often, reset when you have to.

  1. Analyse as built (read-only): one fresh reviewer subagent per architecture area describes how the app actually behaves today, with file:line references, including business logic sitting in UI files.
  2. Map interpretations (read-only): each brief principle is classed as consistent, divergent or contradictory, with the failure mode recorded.
  3. Rewrite the design (you decide): one design call per divergent or contradictory principle, with options, a recommendation, and room for your own answer. The old brief is archived; a human-owned or hook-protected brief gets a proposal file instead.
  4. Plan the rebuild: sized TASKS.md items naming the exact files each may touch. The rebuild runs through /next-task as normal.
  5. /drift-reset postmortem: after the rebuild, checks the app against the recorded failure modes and proposes new check rules as design calls, turning repeated judgment into checks.

Only you start a reset. It writes no application code, uses no destructive git commands, recommends its own branch, and keeps everything in product/resets/YYYY-MM-DD/.

Also in 1.3.0:

  • /align names the conflicts it reports in its commit body, and recommends /drift-reset when a conflict comes back or involves several principles.
  • The reviewer agent also runs drift-reset area analyses, still read-only.
  • rules.md routing lists drift-reset.md and the previously missing refresh-model-sizing.md.
  • selftest.sh checks that every procedure is routed in rules.md, and the dash check now covers new files before they are committed.

Since 1.1.0

  • 1.2.3: Design sessions show the relevant brief sections and decision-log lines before presenting options.
  • 1.2.2: No en or em dashes in the repository; selftest.sh enforces it.
  • 1.2.1: Model sizing aligned: five capability tiers in rules.md, escalation to the next size's model on a failed build, model IDs only in the refresh-model-sizing procedure.
  • 1.2.0: Gitleaks replaces the regex secrets scan, in the check and the pre-commit hook. Credential files are denied to the AI.
  • 1.1.1: Canonical assistant-support wording.
Assets 2
Loading