jabworks-agentic-toolkit

Lean agentic workflow skills: tiered task routing, discovery gate, design-time wireframes and diagrams, lean plan drafting, opt-in tests-first, subagent coordination, end-of-task quality gate, on-request code review, preflight done-check, root-cause-first debugging, cited multi-source research, and feature spec management.

16 skills · 5 agents · hooks

Operate the toolkit repo itself: repo orientation, change control and publish gating, skill-authoring standards, debugging playbook, failure archaeology, plugin schema reference, and research frontier for jabworks/agentic-toolkit.

8 skills

File-based project backlog in committed files: open items in docket/DOCKET.md, closed items archived per year with verification records, ids never reused. The record skill captures and closes items, groom sweeps for stale and ghost work and picks what to do next; a dependency-free CLI, bundled MCP server, and self-contained HTML board keep the mechanics byte-exact.

3 skills · 1 MCP server

Continuous memory for Codex: three hooks share one idempotent rollout sync, so capture is exactly-once regardless of hook order and a hard-killed session is recovered at next start. Memory ages through buffer, daily, recent, and archive tiers; explicitly pinned facts are never auto-compressed.

2 skills

Durable, local usage tracking for Claude Code and Codex: copies session metrics into a private store that survives transcript pruning, with limit snapshots and no prompt text by default.

1 skill · hooks

Use when adapting a generic skill or scaffolding code to match the jabworks stack, conventions, and working preferences.

1 skill

Write code that matches the jabworks house style the first time: enforced rules from the style-guide configs (TypeScript, ESLint/oxlint, Prettier/oxfmt, Stylelint) plus a confidence-marked judgment layer for naming, error handling, React anatomy, monorepo boundaries, and Tailwind.

1 skill

Craft a conventional-commit message from the diff and run the commit safely — review before staging, no blind `git add .`, multiline bodies via -F, branch check first.

1 skill

Decision router for git: given a situation, pick the right operation (undo/discard/stash/merge/push) and run it safely with an undo path. Cross-links to git-commit for committing.

1 skill

Decision router for git worktrees: create an isolated workspace for a task or an agent, list and switch trees, move work between them, prune and remove cleanly, recover a broken one. Native-first — when the host manages worktrees, using git worktree add alongside it creates phantom state it cannot see.

1 skill

Cut releases safely: detects the repo's machinery (AGENTS.md overrides, changesets, this toolkit's plugin tagging, plain GitHub tag + release), always dry-runs the full plan, executes on one confirm, and knows the rollback path for every step.

1 skill

Creates and resumes handoff documents for seamless Claude Code session transitions. Use when context window approaches capacity or a work session is ending.

1 skill

Generate an explorable HTML report of session usage — tokens, cache, subagents, skills, expensive prompts. Works for Claude Code and Codex.

1 skill

Catalog and browse a growing specs/ tree: a plain-markdown index any agent can reference, plus a folder-grouped doc site reusing the plan-review renderer. No external services or plugins.

1 skill