alirezarezvani/claude-skills / engineering/handoff/skills/handoff
handoff
VerifiedCompact the current conversation into a handoff document for another agent to pick up, referencing existing artifacts (PRDs, plans, ADRs, issues, commits, diffs) by path or URL instead of duplicating them. Use whenever the user wants to hand off to a fresh agent, is ending a session mid-task, asks for a handoff doc or a summary of state, decisions and next steps for the next session, or starts a new session that picks up prior work.
| File | Kind | Size |
|---|---|---|
| SKILL.md | skill md | 3.0 KB |
| references/companion_tooling.md | markdown | 3.0 KB |
| references/deduplication_discipline.md | markdown | 6.3 KB |
| references/handoff_structure.md | markdown | 5.9 KB |
| references/next_session_skill_matching.md | markdown | 6.1 KB |
| scripts/artifact_deduplicator.py | script | 8.5 KB |
| scripts/handoff_template_generator.py | script | 7.9 KB |
| scripts/skill_recommender.py | script | 8.1 KB |
references/companion_tooling.md
View on GitHub# Companion Tooling
Handoff-generation tools + cs-* wrapper layered on top of Matt's handoff skill.
## Validation Tools (stdlib Python)
| Tool | Purpose | Run when |
|---|---|---|
| `scripts/handoff_template_generator.py` | Generate a markdown scaffold tailored to next-session focus. Supports `--mktemp` for the path pattern Matt named | Starting a handoff document |
| `scripts/artifact_deduplicator.py` | Detect PRD/ADR/issue/commit content that should be replaced with a reference instead of inlined | Pre-flight check on a handoff draft |
| `scripts/skill_recommender.py` | Match handoff content to skills in this repo, ranked by signal strength | Producing the "Skills to use" section |
All three:
- Stdlib-only
- Run with embedded sample if no input provided
- Output text or JSON (`--output json`)
## The `mktemp` Path Pattern (Matt's Convention)
Matt's SKILL.md specifies:
> "Save it to a path produced by `mktemp -t handoff-XXXXXX.md` (read the file before you write to it)."
`handoff_template_generator.py --mktemp` honors this — uses `tempfile.mkstemp(prefix="handoff-", suffix=".md")` under the hood, returns the path so the caller can read-verify before writing the final content.
## cs-handoff-author Persona Agent
Lives at `../agents/cs-handoff-author.md`. Voice: continuity-focused, no-duplication-tolerated. The persona's hard rule: **if you find yourself typing content from a PRD/plan/ADR/issue, stop and replace with a reference**.
## `/cs-handoff` Slash Command
Lives at `../commands/cs-handoff.md`. Single-trigger handoff with argument hint per Matt's convention: `/cs-handoff <what-next-session-is-for>`.
## Why Wrap Matt's Original
Matt's handoff skill is intentionally minimal (1 paragraph). The wrapper adds:
1. **Tailored templates** — different next-session focuses (deploy/review/debug/design/test) emphasize different sections
2. **Dedup enforcement** — Matt's "do not duplicate" rule, programmatically checked
3. **Skill recommendation** — Matt says "suggest skills to be used" — the recommender automates this from handoff content
## Attribution
Original: [matt-pocock/skills/skills/productivity/handoff](https://github.com/mattpocock/skills/tree/main/skills/productivity/handoff) (MIT).
---
**Source authorities (non-exhaustive):**
- **Matt Pocock — handoff** (https://github.com/mattpocock/skills/, MIT) — the upstream source + mktemp convention + no-duplication rule
- **Anthropic — Multi-agent + session continuity patterns** (https://docs.claude.com/en/docs/agents) — handoff documentation patterns
- **Karpathy, A. — LLM Wiki pattern** (public commentary) — persistent context across sessions
- **Pinker, S. — "Sense of Style"** (2014) — write for the reader who lacks your context
- **Engineering team patterns — Runbook + Playbook discipline** — capturing context for the next on-call engineer
- **DRY principle (Hunt & Thomas, "The Pragmatic Programmer", 1999)** — Don't Repeat Yourself; references > copies
- **GitHub PR description conventions** — what context belongs in handoff vs PR vs ADR
Related skills
| Skill | Stars | Updated | Works with | Status | Install |
|---|---|---|---|---|---|
| reverse-engineer-anythingmorluto/reaReverse engineer native, managed, Electron/JavaScript, packaged, firmware, and browser targets with REA, and build agent environments (AgentEnv bundles) of web applications from a live instance or a capture, and of Linux desktop apps with a task from a request. Use shipped-artifact or requested runtime evidence to explain features, compare versions, decompile code, or guide a reconstruction. Skip REA for ordinary source-repository architecture analysis. | Stars: 472k | Updated | Claude CodeCodex | Verified | Install |
| diagnosing-superpowersobra/superpowersUse when a superpowers session went wrong and your human partner wants to know why — repeated work, ignored plans, stumbles, poor results, a skill that didn't fire, "it took too long", "why is it so expensive", "what is it doing" — or wants to build a bug report for the superpowers maintainers, for the current session or a past one identified by id or path, on any harness. | Stars: 322k | Updated | Claude CodeCodex | Verified | Install |
| dispatching-parallel-agentsobra/superpowersUse when facing 2+ independent tasks that can be worked on without shared state or sequential dependencies | Stars: 322k | Updated | Claude CodeCodex | Verified | Install |
| executing-plansobra/superpowersUse when executing an implementation plan in the current session as the implementer yourself — your human partner chose inline execution, or no subagent tool is available | Stars: 322k | Updated | Claude CodeCodex | Verified | Install |