Progress tracking in project artifacts -- plan, task, and BDD spec documents. Use after completing each phase of any workflow (feature, refactor, bug-fix), at checkpoints during audit passes, when adding or updating behavioral specs, or when the user asks to update project status.
Commit message format rules. Use whenever staging, committing, or describing changes -- including when the user asks to commit, when preparing a PR, or when writing a changelog entry. NOTE: Never commit autonomously. Always prepare a message and ask for user approval before committing.
Ruff and Pyright configuration standards for Python projects. Use when setting up a new Python project, auditing an existing project's tooling config, or applying linting/type-checking fixes. Covers pyproject.toml settings, taskipy tasks, and pre-commit hooks.
Structured workflow for refactoring existing code without changing observable behavior. Use when restructuring abstractions, extracting ports or adapters, converting inheritance to protocols, or adapting callers to new interfaces -- any change where the goal is structural improvement, not new capability. NOT for new features or enhancements (use feature-workflow). NOT for bug fixes.
Tool and terminal decisions for every task. Use for any request involving file reads or edits, running tests, executing searches, git commands, or terminal work -- even when the user does not mention tool choice. Determines whether to use VS Code integrations, MCP tools, or terminal commands.
Read FIRST, before skill-compliance and all other skills. Defines the emission protocol for all skill activity. Every skill loaded in a session must emit skill.invoked immediately after being read. Every completed skill must emit skill.completed. Timestamps are reviewed for compliance.
Read immediately after the telemetry skill. Before performing any task, identify the task type, the files being modified, and the relevant languages or subprojects, then load the right skills and references and confirm them to the user. Use at the start of every task, after telemetry.
Feedback loop procedure for implementing BDD test modules. Use when implementing a spec doc -- covering one test module from spec to type-checker-clean, self-audited output.