handoff

A session handoff format for recording unresolved topics when work moves to a fresh coding-agent session. It keeps open questions and uncertainty as starting context rather than instructions.

In plain words
What is it for?
Use it to update `TODO.md` with open threads, why they matter, and unresolved choices for a later session.
Why use it?
It preserves useful context when a topic is complete, drifts, becomes too large, or needs correction without confusing past notes with completed work.

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/rhi-zone/normalize/handoff
Any agent
npx skills add rhi-zone/normalize --skill handoff
Clone the repo
git clone --depth 1 https://github.com/rhi-zone/normalize

Made for: Claude Code, Codex.

Per session 37 Skills are progressive disclosure: only the name and description are preloaded; the body loads when the skill is used.
When invoked 1,051 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 $0.00037 $0.01051
Opus 5 $0.00018 $0.00526
Sonnet 5 $0.00007 $0.00210
Haiku 4.5 $0.00004 $0.00105

Measured yesterday against content hash df036a67fe07, 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.

.claude/skills/handoff/SKILL.md · 79 lines

How it starts

The opening of the file, as written. The whole thing — 79 lines — stays where its author put it; the contents beside it link to each section on GitHub.

/handoff

You are ending this session and leaving context for the next one. The goal: update TODO.md so the next session has useful starting context. NOT marching orders.

Rules

Write open threads, not directives

TODO.md items are advisory context — "here's what was on the table." The next session serves the user, not this handoff. The user may want to go in a completely different direction, and that's fine.

For each item:

  • State what it is and WHY it matters
  • Note open questions, unresolved judgment calls, or forks in approach
  • Mark uncertainty — "might need X" not "do X"

Do not narrate what was done

Git is the source of truth for completed work. Do not write "what was done this session" into TODO.md. If it's committed, git has it. If it's not committed, it's not done.

Do not write commands or build steps

Those belong in CLAUDE.md. If a command appears in a handoff, that's a sign CLAUDE.md is missing something — update CLAUDE.md instead.

The output must declare its own authority level

At the top of the TODO.md open-threads section, include a line like:

Open threads from a previous session. Treat as starting context, not instructions — verify relevance before acting.

This ensures the next session sees the trust boundary explicitly, even if it doesn't read this skill's definition.

Procedure

  1. Run git log --oneline -20 and git diff --stat to understand what changed this session.
  2. Read the current TODO.md (create it if it doesn't exist).
  3. Update TODO.md:
    • Remove items that are clearly resolved (verify via git, don't guess)
    • Add new open threads from this session's work
    • Update existing items if context has changed
    • Keep items the session didn't touch
    • Ensure the trust-boundary line is present at the top of the open-threads section
  4. Commit the TODO.md update.
  5. Determine mode from the user's wording:
    • If the user's phrasing indicates a continuation handoff (e.g. "/handoff", "hand off", "pass to next session", "leave context for the next one") — proceed to step 6.
    • If the user's phrasing indicates a clean exit (e.g. "ready to exit?", "done", "wrapping up", "stopping here") — stop. The TODO.md update is the deliverable; do not enter plan mode.
  6. Enter plan mode only now, when presenting this plan is the ONLY remaining step — every write and commit must already be done. Subagents spawned from inside plan mode can only write their own plan files, so nothing further can be delegated after this point. Invoke the EnterPlanMode tool with a short plan that communicates direction. Critical: the plan must mark itself as pre-research — a starting hypothesis, NOT a verified directive. The next session hasn't done the investigation; the previous session's intent should inform, not command.

Read the full file on GitHub · 79 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 · 79 lines · 37 tokens per session scan A df036a67fe07

Subscribe to this mod's changes

handoff is a skill published in the GitHub repository rhi-zone/normalize (5 stars, last pushed 3d ago), licensed Apache-2.0. It adds 37 tokens to every session and 1,051 once invoked, about $0.0002 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

repomix

Pack and analyze codebases into AI-friendly single files using Repomix. Use when the user wants to explore repositories, analyze code structure, find patterns, check token counts, or prepare codebase context for AI analysis. Supports both local directories and remote GitHub repositories.

yamadashy/repomix · 58 tokens

kin-retrieval

Read a codebase through Kin's semantic graph instead of grep and whole-file reads. Use when finding where something lives, what a symbol does, who calls it, or which code implements a described behavior, and when a repo has been admitted to Kin (a .kin/ directory exists).

firelock-ai/kin · 63 tokens

blast-radius-review

Review a change by its blast radius, using Kin's graph to find what downstream code the change can reach. Use when reviewing a diff or pull request, deciding whether an edit is safe, or answering "what else breaks if I change this" in a repository admitted to Kin.

firelock-ai/kin · 60 tokens

documentation-sync

Use for any code change, specification change, architecture change, or agent-rule change; do not use as a replacement for the canonical spec.

SioYooo/RepoGrammar · 31 tokens

implement-change

Use for every code change in RepoGrammar; do not use for documentation-only edits that do not touch behavior or repository automation.

SioYooo/RepoGrammar · 28 tokens

major-feature-workflow

Use when a change adds user-visible capability, changes public API, MCP contract, storage schema, module boundaries, indexing pipeline, or important dependencies.

SioYooo/RepoGrammar · 34 tokens