Claude Handoff

Before you start: read this whole install prompt, tell me in a few lines what this skill does and what installing it involves, and wait for my go-ahead before you do anything else. Prerequisites — Framing first, because this one acts at use time: this skill does not only write a handoff summary — when used, it saves that summary of my current conversation to my operating system's temporary directory and immediately launches a fresh background agent (a claude --bg process) seeded with the summary as its prompt, starting in my current working directory, which then begins the work right away and which I manage afterwards with claude agents. Its own instructions require redacting API keys, passwords and personally identifiable information from the summary before writing it, and referencing existing artifacts by path or URL instead of duplicating them — but that redaction is done by the agent running it, so I should treat any sensitive conversation as something to check before handing off. It is an upstream in-progress (beta) skill, not shipped in the author's plugin, and it is user-invoked only. None of that happens during installation: no account, no key, no payment is needed to copy its two files, and no agent is launched, no summary is written, during this install. Install "Claude Handoff" for me. When I invoke it, it writes a summary of our current conversation that a fresh agent can continue from — tailored to the focus I pass as an argument if I give one, with a suggested-skills section naming which skills the next agent should call, and with sensitive information redacted as its instructions require — saves that summary to my operating system's temporary directory, and launches a named background agent seeded with the summary as its prompt (passed through the file so the shell does not expand its contents), starting in the current working directory and returning immediately, for me to manage with claude agents. Its folder is only those instructions plus its agent policy file, which marks it user-invoked only; upstream it sits in the in-progress bucket, public for feedback and not shipped in the plugin. Repository: https://github.com/mattpocock/skills (the skill is the whole skills/in-progress/claude-handoff/ folder at the pinned commit: SKILL.md plus 1 companion files, 2 files in total) Pinned version: install the code at commit 5beb35b040e3de97e54fa9e386d7e42d15c295c2 exactly. Do not substitute a newer version. If this commit is unreachable, stop and tell me. 1. Clone https://github.com/mattpocock/skills into a temporary folder, check out commit 5beb35b040e3de97e54fa9e386d7e42d15c295c2, and copy the skills/in-progress/claude-handoff/ folder from that checkout. Then compare every file in the folder with the fingerprints recorded for this listing on 2026-10-22: [FINGERPRINTS] - SKILL.md: SHA-256 79357a18425d66d4c24048bedf83dceb79d3bba356f38014950f764a7fab4026 - agents/openai.yaml: SHA-256 2308a672978864633286fb9f0792b08657e501426b3f7883fc418d5d2eac5eb0 If a file does not match, it changed since the listing was prepared: do not install it yet, tell me what you found and let me decide. 2. Read SKILL.md in full, and tell me in plain words: what the skill does when used, what it connects to, whether it reads or sends any personal information or credentials, and anything that seems unusual. Stop on any red flag and tell me before going further. Confirm these points from the files: - The skill folder holds 2 files only: a 1,424-byte SKILL.md with the handoff workflow above, and agents/openai.yaml declaring its display name and its user-only invocation policy. There is no script and no other code in it; the background-agent launch it describes is a command the agent runs at use time, not an installed program. - Nothing is sent anywhere during installation itself; the only download at install time is the skill folder from GitHub at the pinned commit. No summary of any conversation is written, and no background agent is launched, during installation. - At use time, what leaves the current conversation is the handoff summary itself: it becomes the prompt of a fresh background agent running on my own machine, in my current working directory. The skill's own text requires that summary to be redacted (API keys, passwords, personal information) and to reference existing artifacts instead of duplicating them — disclosures repeated in the prerequisites above. - Do not write any handoff summary, do not launch any background agent, and do not use the skill in any way during this install: using it is a separate step that only starts when I invoke it later. 3. When I say "yes, install", install it as a skill: copy the whole claude-handoff folder into the agent's skills directory, in a folder named exactly "claude-handoff", keeping its 2 files in their original relative layout (SKILL.md at its root). 4. Verify with no network calls: frontmatter valid (name: claude-handoff), the 2 files in place with matching fingerprints. Do not run anything from the folder and do not use the skill to verify. 5. Report what was installed, where, and what I still need to do myself (for example: that it is an upstream in-progress beta skill, that using it writes a redacted summary of my conversation to my temp directory and immediately launches a background agent from it, and that none of that happens at install time — using it is a separate step that only starts when I invoke it). GitHub is optional: if I have a GitHub account or the gh CLI, you may use it; otherwise public access is fine. Never require it unless it's in the prerequisites above. Rules: don't touch anything outside the temp folder and the install target. If anything looks off, stop and ask me.