handoff

A tool for writing and checking session handoff documents. These documents tell the next developer what work is complete, what remains, and how to continue or review it.

In plain words
What is it for?
Use it to update a workstream’s continuation document, create an external review record, validate the document, and optionally add review checkpoints to a roadmap.
Why use it?
It prevents important project context from being lost between coding sessions. It also keeps continuation notes current and separates them from longer-term lessons or milestone tracking.

Skill for Claude CodeCodex

Install

Getting it into your agent

One page per mod, every tool's command on it. A separate URL per tool would split the same page into five that compete with each other.

agentmods
npx agentmods add skills/chris-dare-dev/agent-kit/handoff
Any agent
npx skills add chris-dare-dev/agent-kit --skill handoff
Clone the repo
git clone --depth 1 https://github.com/chris-dare-dev/agent-kit

Made for: Claude Code, Codex.

Per session 187 Skills are progressive disclosure: only the name and description are preloaded; the body loads when the skill is used.
When invoked 2,791 The whole file, excluding the scripts and references it only reads on demand.
Security scan A 0 findings. Scan, not verified.
Origin original No closer match found in the catalogue.
Token cost

What it costs to keep this loaded

Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.

ModelPer sessionOnce invoked
Fable 5 $0.00187 $0.02791
Opus 5 $0.00093 $0.01396
Sonnet 5 $0.00037 $0.00558
Haiku 4.5 $0.00019 $0.00279

Measured yesterday against content hash db42e8baebb0, method: parsed. Prices are Anthropic first-party input rates as of 2026-08-30, from the pricing page.

Security

Grade A, and why

handoff scanned grade A with 0 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured yesterday.

A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.

Nothing flagged

None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.

data/skills/handoff/SKILL.md · 201 lines

How it starts

The opening of the file, as written. The whole thing — 201 lines — stays where its author put it; the contents beside it link to each section on GitHub.

Handoff

Author continuation and/or review handoffs against the canonical contract in .claude/references/handoff-contract.md, validate them with .claude/scripts/handoff-validate.py, and wire review handoffs into the roadmap as optional audit checkpoints. Works identically from Claude Code, Codex, and OpenCode — the contract and validator are provider-neutral; the Obsidian surfacing step degrades gracefully on machines without the vault tooling.

Continuations are live documents, not snapshots. When the workstream store is available (contract §2a, workstream-state.md), the continuation handoff is Workstreams/<slug>/continuation.md — one mutable file per workstream, gist-backed so every machine sees the same state — and this skill UPDATES it in place rather than minting a dated file. Review handoffs stay dated snapshots in the same workstream directory. The legacy plans/HANDOFF-<date>-... form is the fallback for machines without the store.

When NOT to use

  • To record a durable lesson or preference — that's /memory-sync / memory, not a handoff.
  • To advance or close a milestone — /milestone-pipeline owns execution state; a handoff only describes it.
  • Mid-task context notes for yourself — handoffs are session-boundary artifacts.

Arguments

/handoff [continuation|review|both] [project-slug] [free-text notes]

  • kind — omit to decide from the session: shipped milestone-scale work (commits landed across ≥1 milestone) → both; mid-stream or light session → continuation only. Say which you chose and why in the final report.
  • project-slug — omit to infer from the session's roadmap/milestone ids. This is the program slug (e.g. identity-seed, svcreg-supplychain-trust), used in the filename, project: frontmatter, and roadmap lookup.
  • notes — anything the user wants emphasized (e.g. "reviewer should focus on the live flip").

Steps

  1. Read the contract first: .claude/references/handoff-contract.md — filename grammar, frontmatter schema, both body templates, the review-checkpoint format. Do not improvise frontmatter keys; legacy variants (type: session-handoff, handoff-kind:) break the vault Bases.

Read the full file on GitHub · 201 lines

Changes

What this file has done since we first saw it

Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.

  1. yesterday First seen · 201 lines · 187 tokens per session scan A db42e8baebb0

Subscribe to this mod's changes

handoff is a skill published in the GitHub repository chris-dare-dev/agent-kit (0 stars, last pushed 24d ago), licensed MIT. It adds 187 tokens to every session and 2,791 once invoked, about $0.0009 per session on Opus 5. A static security scan graded it A with 0 findings. No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-31.

Related

Other skills, from other repositories

ask-user

You MUST use this before high-stakes architectural decisions, irreversible changes, or when requirements are ambiguous. Runs a decision handshake with the askuser tool: summarize context, present structured options, collect explicit user choice, then proceed.

edlsh/pi-ask-user · 48 tokens

kiwix

Search and read articles from a local Kiwix server — offline Wikipedia, documentation, Stack Exchange, and other ZIM content. Use when the user asks to look something up, find documentation, or read an article from local knowledge bases.

OscillateLabsLLC/kiwix-mcp · 52 tokens

algolia-autocomplete

Building autocomplete and search-suggestion experiences with the Algolia Autocomplete UI library — distinct from InstantSearch. Covers query suggestions, multi-source autocomplete (suggestions + recent searches + records + redirects + content), the plugin architecture (@algolia/autocomplete-plugin-), keyboard nav…

bpainter/composable-dxp-claude-marketplace · 131 tokens

algolia-contentful-integration

The Contentful ↔ Algolia integration we ship by default on Composable DXP engagements — choosing between the Marketplace app, the Ingestion API source connector, and a custom Vercel-Function indexer; mapping Topics & Assemblies to Algolia records; per-locale fan-out; preview vs. delivery indices; webhook signature…

bpainter/composable-dxp-claude-marketplace · 153 tokens

algolia-index-design

Index and record design for Algolia — designing the JSON record shape that powers search, deciding searchable vs. faceted vs. displayed attributes, choosing per-type indices vs. one federated index, structuring distinct/grouping for variant collapse, and architecting replicas (standard vs. virtual) for sort orders.…

bpainter/composable-dxp-claude-marketplace · 139 tokens

algolia-indexing-pipeline

Getting data into Algolia — full reindex vs. partial updates, atomic replacement via replaceAllObjects, batch sizing, idempotency, streaming pipelines (webhook-driven incremental sync), backfills, the Ingestion API for managed source connectors, and the Slalom default pattern of Vercel Functions + Contentful webhooks…

bpainter/composable-dxp-claude-marketplace · 122 tokens