skills/mctx-init/SKILL.md
Bootstrap a project to use mctx by injecting a managed block into AGENTS.md. Tool-agnostic guidance every agent reads. Runs `mctx init` and shows the diff. Use when starting a new project, onboarding a repo, or asked to "init mctx" / "add mctx guidance" / "set up AGENTS.md".
npx skillsauth add tmsjngx0/mindcontext-core mctx-initInstall this skill globally with one command. Works with Claude Code, Cursor, and Windsurf.
3 of 9 scanners reported clean
Some scanners were skipped, did not run, or reported a non-clean status. Review each row below.
Thin wrapper around the mctx init CLI command. The CLI is the real interface;
this skill just makes it discoverable via /mctx-init and shows the result.
Injects a managed block (<!-- MCTX:START --> ... <!-- MCTX:END -->) into
AGENTS.md describing:
mctx issue list|show|create|close)mctx catchup / mctx handoff)--force: overwrites AGENTS.md entirely with just the managed block.mctx init with any passed flags (e.g. --force).AGENTS.md (use git diff AGENTS.md if tracked,
otherwise cat AGENTS.md).Because Claude Code users can invoke /mctx-init and get the same bootstrap
experience as running the CLI manually. Users of Codex, Cursor, or any other
agent just run mctx init directly — the CLI is the source of truth.
development
Test-Driven Development workflow following Kent Beck's methodology. Auto-loads when user asks about "tdd", "test driven", "red green refactor", "write tests first".
testing
Analyze current session and generate a self-contained context prompt for the next Claude session. Includes project context, decisions, next steps, and tech stack. Use when ending a work session.
tools
Summarize current project state after a long break. Analyzes recent git history, beads issues, and memory to produce a one-page orientation. Use when returning to a project cold, or when asking "what was I doing", "where did I leave off".
data-ai
Load context from vault memory. Temporal queries (yesterday, last week, session history) use native JSONL timeline. Topic queries use QMD BM25 search. "recall graph" generates interactive temporal graph of sessions and files. Every recall ends with "One Thing" - the single highest-leverage next action synthesized from results. Use when user says "recall", "what did we work on", "load context about", "remember when we", "prime context", "yesterday", "what was I doing", "last week", "session history", "recall graph", "session graph".