Use when building a managed team skills library for a real stack. Map work to shelves, browse before curating, write meaningful `whyHere` notes, and create a starter pack once the first pass is solid.

Writing effective code documentation - API docs, README files, inline comments, and technical guides. Use for documenting codebases, APIs, or writing developer guides.

Use when regenerating README.md and WORK_AREAS.md in a managed library workspace. Always dry-run first to preview changes.

Use when exploring the ai-agent-skills catalog to find, compare, and evaluate skills before installing. Always use --fields to limit output size and --dry-run before committing to an install.

Backend API design, database architecture, microservices patterns, and test-driven development. Use for designing APIs, database schemas, or backend system architecture.

Use when checking the overall health of a skills library. Run doctor, validate, check for stale skills, and verify generated docs are in sync.

Clarify requirements before implementing. Do not use automatically, only when invoked explicitly.

Clear due memory reviews with free recall — the two-minute habit that makes learning permanent. Use when reviews are due, or the user wants to review, practice, or "do my engram reviews".

Learning telemetry, strategy, and schedule — retention stats, calibration, grader audit, n-of-1 experiments, HTML dashboard. Use for "how am I doing", weekly check-ins, strategy questions, auditing the grader, or adjusting how Engram teaches.

Analyze product and technical requirements for the PigeonPod project with software engineering rigor. Use when users ask to evaluate a feature, enhancement, non-functional requirement, integration, or migration for value, feasibility, architecture fit, implementation impact, risk, delivery scope, or tradeoffs. Do not use for bug triage or root-cause analysis; use `bug-analysis` for bugfix-oriented work. Always inspect current repository docs and code first, then use MCP tools including Context7 to verify external library, framework, or API constraints before concluding.

Draft, refine, and publish bilingual PigeonPod GitHub release notes from commits on the `release` branch. Use when Codex needs to compare commits since the latest published GitHub release, write a new local release note under `dev-docs/release-notes`, align English and Chinese release-note sections after user edits, or create/update a GitHub Release while keeping the markdown H1 as the GitHub release title instead of the release body.

Find open GitHub issues that are covered by a specific release note, draft issue replies in the issue author's language, post approved comments with `gh issue comment`, and recommend whether each issue should be closed. Use when Codex needs to turn a shipped release into structured GitHub issue follow-up, especially for PigeonPod release-note-driven maintainer workflows.

Review PigeonPod GitHub issues end to end. Use when the user asks what an issue means, whether it is valid, how to reply, whether to add it to the GitHub Project, or to turn it into a tracked task. Read the issue and comments, inspect relevant local docs and code, explain the real requirement or bug precisely, draft a maintainer reply, and only after explicit approval perform GitHub writes. When creating a task, first recommend `Priority`, `Size`, and `Estimate`, wait for maintainer confirmation or overrides, then write those values into the project task fields.

Generate and process the daily PigeonPod issue triage report. Use when reviewing open GitHub issues incrementally with a text cursor, classifying issues as requirement, bug, or discussion, drafting maintainer-facing analysis, or executing developer-approved follow-up actions from the daily report.

Analyze software bugs for the PigeonPod project with a bugfix-first workflow. Use when users report broken behavior, regressions, incorrect results, crashes, data inconsistencies, sync/download failures, or ask for root-cause analysis, fix strategy, repro analysis, severity assessment, or regression-risk evaluation. Read current repository docs and code first, then use MCP tools including Context7 only when framework, library, API, or external-service behavior must be verified.

Performs placebo-in-time sensitivity analysis with hierarchical null model and optional Bayesian assurance. Use when checking model robustness, verifying lack of pre-intervention effects, or estimating study power.

Fit, summarize, plot, and interpret a chosen CausalPy experiment. Use after the causal method has been selected, including when configuring PyMC/sklearn models and scale-aware custom priors.

Load built-in CausalPy example datasets for demos, tutorials, tests, and quick causal-analysis prototypes. Use when the user needs sample data or asks which demo datasets are available.

Challenge causal claims through structured threat assessment, counterfactual reasoning, and CausalPy falsification checks. Use when validating whether a causal effect is real or when the user asks "is this effect real?" or "can I trust this result?"

Interactive development in marimo notebooks with validation loops. Use for creating/editing marimo notebooks and verifying execution.