
Claude Skills by Catherine1401
github.com/Catherine1401Prepare focused Git commits when asked to commit or split commits. Never stage unrelated changes or run git commit before explicit user approval.
Create or update an agent skill with appropriately scoped instructions and necessary resources. Use for skills; not policies, plugins, or project implementation.
Update the shared user-level policy for Codex, Claude Code, and Cursor, then install it. Use for universal user instructions; not project policies, skills, sub-agents, or plugins.
Test fixture for manifest-driven installation.
Create or update a searchable task context handoff when the user asks to report, save context, hand off work, or prepare another agent to continue.
Create or update an agent skill scoped to a single project's own skill directory (.claude/skills or .codex/skills), not this repo's portable skills. Use for project-local skills; not portable skills, policies, plugins, or project implementation.
Capture useful lessons from the current session as clear, reusable notes for the user. Use when asked to learn from a session or save what was learned.
Verify changes against user-level and project-level AGENTS.md and CLAUDE.md rules through an isolated subagent, only when the user explicitly invokes this skill.
Give blanket approval to the agent's latest pending confirmation request, or confirm the agent's latest result worked when it asked the user to verify. Use when the user invokes /ok.
After writing or changing code, guide the user through testing the changed behavior. Provide runnable mocks when the user explicitly says required APIs or backend dependencies are unavailable or unfinished.
Update the current project's own policy file for a project-specific requirement. Use for rules that apply only to this project; not user-level policy (use user-policy), skills, or plugins.
Read a video file (.mp4/.mov/.webm) by extracting frames with ffmpeg and summarizing what it shows, e.g. a QA bug recording. Use when given a video path; not for still images or live app testing.