Handoff by caio
Production-grade AI skills for Claude Projects and Cowork. Built by Stefanos Karagos at CAIO Group (wearecaio.com).
npx -y skills add karagos/ai-skills-by-caio --skill handoff-by-caioAssembled from the repository path, not quoted from the project. Check it against their README if it does not work.
2 things to look at
- no licenseNo license file was found in the repository. Code published without one is not open source by default, so using it at work is a question for whoever answers licensing questions where you are.
- 2 stars2 stars. Stars are a popularity signal and not a quality one, but at this level it is likely that nobody has read this closely except its author, and you would be relying on your own review.
What its author says it does
Copied from the file, not written here
Generates a structured Markdown handoff file to move an active conversation into a new chat, preserving full context, decisions, artifacts, and a ready-to-paste resume prompt. ALWAYS trigger this skill when the user says "move to a new chat", "move this to a new chat", "handoff", "/handoff", "continue in a new chat", "let's move", "new conversation", or any variation indicating they want to transfer the current session to a fresh Claude conversation. Do NOT ask for clarification — execute immediately and produce the .md file.
SKILL.md
6.3 KB, as published. Nobody here has run it
Handoff Skill
by CAIO Group — wearecaio.com
Move any Claude conversation to a new chat with zero re-explanation.
Generates a structured .md file with full context, classified artifacts, and a paste-ready resume prompt.
Trigger Phrases
Execute immediately (no confirmation needed) on any of these:
/handoff- "move to a new chat"
- "move this to a new chat"
- "continue in a new chat"
- "let's continue in a new conversation"
- "handoff file"
- "new chat" (when context implies session transfer)
Execution Steps
Step 1: Analyze the conversation
Scan the full conversation history and extract:
- Session type: What kind of work was this? (e.g., proposal, slide deck, automation, strategy, training, code, analysis, research, writing)
- Core objective: What was the user trying to accomplish?
- Decisions made: Key choices, directions taken, things approved or rejected
- Current state: What is built, written, or resolved vs. what is still open
- Artifacts created: Any files, code blocks, documents, scripts, dashboards, or structured outputs produced
- Next steps: What was the natural next action before the session ended?
Step 2: Classify artifacts
For each artifact identified, assign one of three tags:
[CARRY OVER]— Essential to continue. Include full content as a code block.[REFERENCE]— Useful context but not required. Include a summary only.[ARCHIVE]— Superseded or no longer relevant. Note it exists, exclude content.
Default rule: When in doubt, tag as [CARRY OVER].
Step 3: Generate the suggested chat filename
Derive a suggested name for the new chat using this format:
[PREFIX] — [Topic] [Qualifier]
Prefix by session type:
| Session Type | Prefix |
|---|---|
| Proposal / Pitch | PROP |
| Slide Deck / Presentation | DECK |
| Automation / Script / Code | AUTO |
| Strategy / Planning | STRAT |
| Training / Education | TRAIN |
| Analysis / Research | ANALY |
| Skill / Tooling | SKILL |
| Writing / Content | WRITE |
| Design / Creative | DESIGN |
| Communication / Outreach | COMM |
| General / Mixed | WORK |
Topic: 2-4 words from the session objective. Title case. No filler words.
Qualifier (optional): Short version or context indicator (e.g., v2, Draft, Client Name).
Examples:
PROP — SaaS Pricing Overhaul v2AUTO — Email Automation PipelineSTRAT — Q3 Market ExpansionSKILL — Handoff Workflow Builder
Step 4: Generate the handoff file
Produce a single Markdown file using the template below. Fill every section. No placeholders in output.
Handoff File Template
# Handoff: [Session Title — descriptive, not generic]
**Date:** [today's date]
**Session Type:** [detected session type]
**Suggested chat name:** `[PREFIX — Topic Qualifier]`
---
## What We Were Working On
[2-4 sentences. What was the task, why it mattered, what approach was being taken.]
---
## Decisions Made
- [Decision 1 — include reasoning if discussed]
- [Decision 2]
- [Decision 3]
---
## Current State
### Done
- [Completed items]
### In Progress
- [Partially done items]
### Not Started
- [Scoped but untouched]
---
## Artifacts
### CARRY OVER — Bring These Into the New Chat
#### [Artifact Name]
**Type:** [code / markdown / config / script / document / etc.]
**Why it's needed:** [one sentence]
[Full artifact content as fenced code block]
---
### REFERENCE — Useful Context
#### [Artifact Name]
**Summary:** [2-3 sentences]
---
### ARCHIVE — No Longer Needed
- [Artifact name]: [one-line reason]
---
## Open Questions
- [ ] [Unresolved decision or pending input]
- [ ] [Ambiguity to address in new chat]
---
## Next Steps
1. [First action — executable immediately in the new chat]
2. [Second action]
3. [Third action]
---
## Resume Prompt
Paste this as your first message in the new chat:
[3-5 sentences in first person that:
- State the session type and goal
- Reference the most critical current state item
- Point to the most important CARRY OVER artifact
- State exactly what the next action is
Must be self-contained: a fresh Claude instance begins work immediately.]
Step 5: Output the file
- Save the completed handoff file to
/mnt/user-data/outputs/handoff-[topic]-[date].md(If this path is unavailable, save to the current working directory) - Deliver the file to the user
- Print a 4-line summary in chat:
- Session type: [detected type]
- Artifacts: [count and classifications]
- First next step: [specific action]
- Suggested chat name:
[PREFIX — Topic Qualifier]
Do not print the full file contents in chat.
Quality Standards
- Every section filled. No placeholder text in output.
- Resume Prompt is paste-ready: no blanks, no angle brackets, no instructions to "fill in X".
- CARRY OVER artifact content is complete, never truncated.
- Decisions reflect actual choices made in the conversation, not generic statements.
- Next Steps are specific enough to act on immediately.
- The output file contains no tool branding, no skill credits, no advertisements.
Edge Cases
No artifacts in conversation: Skip the Artifacts section entirely.
Short conversation (under 5 exchanges): Collapse "Done / In Progress / Not Started" into a single "Current State" paragraph.
Multiple unrelated topics in one conversation: Generate one handoff file per topic thread, named distinctly. Inform the user that multiple files were created.
Sensitive content (names, financials, credentials): Include as-is. The user controls their own handoff files.
Output path unavailable: Save to current working directory and inform the user of the actual file path.
Gives 0 of the 12 instructions most agent orchestration skills give
Counted across 742 of the 995 authors here whose files we hold, read 2026-08-06
- run the full test suite after integrating changesin 53 of 742, across 20 files
- reference existing artifacts by path or URLin 52 of 742, across 22 files
- dispatch one agent per independent problem domainin 50 of 742, across 17 files
- verify fixes do not conflictin 45 of 742, across 13 files
- include a suggested skills section in the documentin 45 of 742, across 15 files
- redact sensitive informationin 41 of 742, across 11 files
- save to the temporary directory of the operating systemin 39 of 742, across 9 files
- tailor the document to user-provided focus argumentsin 39 of 742, across 9 files
- spot check agent changes for systematic errorsin 34 of 742, across 7 files
- write a handoff document summarising the current conversationin 31 of 742, across 6 files
- assign each agent a specific scopein 23 of 742, across 8 files
- provide specific scope and clear goalin 23 of 742, across 5 files
Said here and by no other author read
- execute immediately without asking for confirmation
- analyze the conversation history
- extract session type, objective, decisions, and current state
- tag artifacts as carry over, reference, or archive
- include full content for carry over artifacts
- generate a suggested chat name
Grouped from the skills themselves: near-identical wordings counted once, and counted by distinct author, so one author publishing three of these counts once.