alfred

Run a deep security and quality audit of the project, present findings, and offer to fix them.

Set up a persona-tailored AI-assisted development environment. This command detects what exists, asks targeted questions, selects a persona, generates your CLAUDE.md, and initializes progressive skill building.

Autonomously fix CI failures in a loop until all checks pass or a stop condition is reached.

Preview, contribute, ingest, or manage encrypted anonymized learning signals.

If `coding_level == "beginner"` in `.claude/.onboarding-state.json`, preface the commit workflow with a brief explanation:

Inventory all result artifacts and format into a summary with full provenance.

Guide new users through checking for a GitHub account, creating one if needed, authenticating, and creating a repository for their project. This command assumes the user knows nothing about GitHub and explains every step in plain language.

Assess the current project's AI-assisted development maturity and recommend improvements.

Use this at the start of any session that involves code changes. Enforces branch hygiene and scopes the session before touching any files.

View, modify, or reset your Alfred persona. Your persona evolves over time based on corrections and preferences, adapting how Alfred explains things and how autonomously it acts.

Manage consent for all Alfred anonymized data collection (telemetry + collective learning signals).

Delete your pilot telemetry and feedback data.

Submit manual feedback about your Alfred experience. Feedback is PII-scrubbed and saved to `.pilot/feedback/`.

Standardized branch-to-PR workflow. Prevents pushing to main and enforces lint checks.

Enables test-driven refactoring with automatic rollback on test failure. Each change is isolated, tested, and committed atomically. If tests fail, the change is rolled back and an alternative approach is attempted.

Analyze accumulated feedback memories and session patterns, then propose promotions to CLAUDE.md rules or hooks. This is the on-demand promotion engine for the self-improvement loop.

Show onboarding progress, graduated habits, and contextual next steps.

Teach development habits one at a time, using domain-specific analogies. Each lesson has 4 phases: Context, Demo, Install, Verify.

Pressure-test the current plan, proposal, or implementation strategy before committing to it.

Use when the user wants to preview or contribute anonymized learning signals to the collective

Use during self-improve to automatically evolve the user persona based on accumulated corrections and usage patterns

Use when the user wants to view, modify, or reset their Alfred persona, or when running /persona

Contextual recommendations for every Alfred command — when to suggest, what to say, and why

Use when starting any session with Alfred installed to orient the user and check onboarding state

6 hooks across 4 events