agentic-toolkit

Use this skill when the user wants review or approval feedback on existing code changes: a PR, diff, patch, workflow file, migration, Terraform/IaC change, dependency bump, auth/security refactor, API change, or SKILL.md update. Trigger on intents like: review, sanity check, LGTM, pre-merge feedback, "before I approve", "before we merge", "any issues with this PR?", "does this look right?", or when a PR link/number is shared for comments. Covers correctness, security, tests, maintainability, API breaking changes, DB migration rollback/locking risk, CI/CD gate bypasses and secret exposure, Terraform least-privilege, risky dependency upgrades, and skill naming/description conventions. Outputs: Blocker, Important, Nit. Do not use for writing or refactoring code, implementing features, debugging errors, creating issues, or generating release notes when the user explicitly wants notes only.

Python coding standards, that covers repository-first conventions, type annotations, error handling, testing, logging, documentation, async code, and dependency hygiene. Activate this skill whenever the user is editing or creating .py files, working in a Python project, or asks for help writing functions, classes, tests, modules, or fixing code - even if they do not say "Python" explicitly. Also triggers on questions about coding conventions, type hints, pytest, imports, logging setup, or dependency management in a Python context.

Creates a GitHub issue from a natural language prompt. Discovers repo templates, fills the best match, posts via gh CLI. Use this skill whenever the user wants to create an issue, open a bug report, file a feature request, submit a ticket, log a bug, report a problem, track something as an issue, or add a ticket — even if they don't say "issue" explicitly.

Always-active response formatting rules. Enforces conciseness, no filler, structured output. Suspends on explicit verbosity requests. Another skill's format takes precedence.