aicontext

Generate project documentation from code and AIContext artifacts — reference (internal technical), guide (user-facing), or both

Use when the user wants project documentation generated from code and AIContext artifacts — reference (internal technical), guide (user-facing), or both.

Draft a GitHub issue based on a discussion or requirements in the current conversation

Draft a pull request title and body based on the current task file and git changes

Fix failing CI checks on the current PR — fetches failures via gh CLI, diagnoses, fixes, pushes, waits for green. Covers lint, type, build, and tests. Retries up to 3 times.

Check unresolved PR review comments and summarize findings

Automate the PR review cycle — fetches review comments, triages, fixes real issues, auto-resolves false positives, runs tests, commits, pushes, then waits for re-review. Repeats up to 5 cycles.

Load task context, show progress, and surface ambiguities and conflicts before resuming implementation

Validate the current task — plan quality, spec coverage, deliverable alignment, and context consistency

Use when the user wants to resume a task — loads context, shows progress, and surfaces ambiguities and conflicts.

Use when the user wants the current task validated — plan quality, spec coverage, deliverable alignment, and context consistency.

Archive completed tasks, specs, and worklog entries — clean up .aicontext/ clutter

Archive completed tasks, specs, and worklog entries — clean up .aicontext/ clutter.

Update all context files (task, spec, task-context, worklog) to reflect current state of work — fixes stale entries, fills missing knowledge, reports what changed

Execute all pending steps in the current task file — creates task-context, checks commit config, runs each step with review-fix inner loop, commits per config

Use when you want to update all context files (task, spec, task-context, worklog) to reflect current state of work. Fixes stale entries, fills missing knowledge, reports what changed.

Use when a task is complete. Verifies all steps are done, applies any task-context Decision Overrides to the spec, fills completion notes, updates changelog, and handles git (commit / push / PR per config).

Use when you want to execute all pending steps in the current task file. Creates task-context, checks commit config, runs each step with a review-fix inner loop, commits per config.

Confirm project readiness in one sentence

Capture a deferred idea to the worklog Ideas section — quickly log anything worth revisiting without losing it mid-session

Add a new step to the current task based on conversation context

Show AIContext quickstart guide with setup, workflows, and best practices

Show all available AIContext skills grouped by workflow stage

Generate missing angles, better implementations, and new combinations — creative expansion that complements /challenge

Pause and push back — surface questions, wrong assumptions, concerns, and better alternatives before proceeding

Commit current changes following project commit rules (template, body settings).

Create a task file from current discussion — lighter alternative to /start-feature

Comprehensive code review — architecture + correctness. Scope with args — diff (default), branch, all, path, or IDE selection. Delegates to reviewer agent for large scope.

Turn the current discussion into a task step and implement it — adds step to task file, updates spec if needed, then implements with review-fix loop

Close out the current task — verify all steps done, sync spec, fill completion notes, update worklog, and handle git (commit / push / PR per config)

Run a structured breadth-first interview to explore a topic thoroughly — walks dimensions, batches independent questions, recommends answers, captures decisions

Mark previous step as done in the task file, then start the next step

Read an existing spec and propose a task breakdown — assesses separable work streams, proposes split, creates task files linked to the spec

Review code changes for bugs, security, edge cases. Scope with args — diff (default), branch, commit, path, or IDE selection. Delegates to reviewer agent for large scope.

Execute a single step from the task plan. Accepts optional step number arg. Asks to commit after completion.

Run a structured discovery interview before starting a new feature — explores the codebase, asks one question at a time, presents decision packages with options + recommendation, then creates a spec and task file

Quick honest take — share thoughts, concerns, or better ideas about the current discussion without a heavy audit

Browser-based investigation using playwright-cli — open pages, inspect elements, check console errors, capture screenshots. Headed mode by default.

Run a structured breadth-first interview to explore a topic thoroughly — walks dimensions, batches independent questions, recommends answers, captures decisions.

Capture a deferred idea to the worklog Ideas section — quickly log anything worth revisiting without losing it mid-session.