CommunityKunst & Designgithub.com

netresearch/jujutsu-workflow-skill

Agent-safe version control with Jujutsu (jj) — jj for local change management, Git as the canonical remote/PR/CI/audit interface. Netresearch agent skill.

Funktioniert mitClaude Code~Codex CLI~CursorAntigravity
npx skills add netresearch/jujutsu-workflow-skill

Ask in your favorite AI

Open a new chat with this agent skill pre-loaded.

Dokumentation

Jujutsu Workflow

For agentic coding, prefer jj (Jujutsu) over raw git whenever a .jj/ repo is present — it is the superior local change layer. Git stays the canonical remote/PR/CI/audit interface: mutate with jj, verify with read-only Git.

1. Detect & gate first

Run ${CLAUDE_SKILL_DIR}/scripts/detect_jj_state.sh, or check manually:

  • .jj/ present → jj repo: use jj, never mutating raw git.
  • .jj/ and .git/ → colocated: mutate with jj, read with git, never touch the git index/staging.
  • only .git/ → plain Git repo: do not introduce jj unless asked.

See references/git-interop.md.

2. Agent-safety rules (non-negotiable)

  • Always --no-pager on output commands; set jj config set --user ui.paginate never.
  • Always -m. Never run editor/TUI forms — bare jj describe|commit|squash, jj split (interactive), jj squash -i, jj resolve, jj diffedit — they hang agents.
  • jj snapshots the working copy only when a jj command runs, not on every file write.

See references/agent-safety.md.

3. Edit loop

jj --no-pager status
# edit files (snapshotted on the next jj command)
jj --no-pager diff
jj describe -m "<project-conventional message>"
jj new -m "<next unit>"     # one change per logical unit — avoids one fat commit

Split a mixed change non-interactively: jj split <path> -m "<msg>". See references/command-map.md.

4. Recover (reversible)

jj --no-pager op logjj undo (last op) or jj op restore <id>. Conflicts are first-class: jj status flags them; resolve by editing markers, then verify. See references/recovery-playbook.md.

5. Hand off via Git

jj git fetch
jj rebase -d <default-branch>
jj bookmark create <branch> -r @-
jj git push --bookmark <branch>      # new bookmarks push directly

Never push to a protected/default branch; never rewrite public history unless allowed. Open the PR with gh/glab. See references/pr-handoff.md.

6. Parallel agents

One jj workspace per concurrent agent — never share a working copy. See references/parallel-agents.md.

7. Verify before "done"

Run ${CLAUDE_SKILL_DIR}/scripts/verify_handoff.sh, or report jj --no-pager status, jj --no-pager log --limit 10, jj --no-pager diff --stat, git status --short --branch. Report exact commands and output; never claim "done/tested/ready" without it; disclose force-pushes, recoveries, conflicts.

Why jj beats git for agents (and when NOT to use it): references/why-jj-for-agents.md.

Verwandte Skills

wondelai/release-it

Build production-ready systems with stability patterns: circuit breakers, bulkheads, timeouts, and retry logic. Use when the user mentions "production outage", "circuit breaker", "timeout strategy", "deployment pipeline", "chaos engineering", "bulkhead pattern", "retry with backoff", or "health checks". Also trigger when designing resilient microservices, planning zero-downtime deployments, or investigating cascading failure scenarios. Covers capacity planning, health checks, and anti-fragility patterns. For data systems, see ddia-systems. For system architecture, see system-design.

community

perasyudha/skill-web3

A secure, non-custodial Web3 & EVM operations skill for AI Agents (OpenClaw, Claude, Cursor) with built-in MCP server, PnL tracking, security audits, and trade monitoring.

community

LcpMarvel/feishu-whiteboard-pro

A Claude Code / agent skill for building genuinely designed, editable Feishu / Lark (飞书) whiteboards — deliberate composition, real hierarchy, a gated pipeline with pre-render fit-check and independent design critique.

community

Laith0003/ux-skill

Design intelligence engine for AI coding tools (Claude Code, Cursor, Windsurf). Deterministic anti-AI-slop linter with 152 rules, 160 brand specs, a 7-axis synthesizer, an MCP server (18 tools), 25 commands, 17 IDEs. Offline, never calls an LLM. MIT.

community

valllabh/am-sandbox-sdk

Shared sandbox SDK: ManagerClient, skills loader, monitor (heartbeat + watchdog + artifact finalize). Consumed by am-pi-agent and am-opencode-agent as a submodule.

community

qodex-ai/multi-agent-orchestration

Design and coordinate multi-agent systems where specialized agents work together to solve complex problems. Covers agent communication, task delegation, workflow orchestration, and result aggregation. Use when building coordinated agent teams, complex workflows, or systems requiring specialized expertise across domains.

community