generate-handover
Skill60 starsChanged 34 days agoMITView on GitHub
Generate a session handover document that captures progress, decisions, and context for seamless session continuity. This documentation shall serve Agents and Humans when working in consecutive sessions with the project. Use this skill at the end of a work session or when context transfer to a new session is needed.
In this repository17 other files it gives its agents.
- AGENTS.mdAGENTS.md
- docs/agents.mdAGENTS.md
- archive-legacy-docsSkill
- author-and-verify-implementation-planSkill
- browser-walkthroughSkill
- create-planSkill
- delegate-analysisSkill
- execute-work-packageSkill
- generate-docsSkill
- report-environment-issueSkill
- resume-planSkill
- review-fixSkill
- review-implementation-planSkill
- review-implementationSkill
- review-planSkill
- update-docsSkill
- update-planSkill
The file
Show its rules and text
We split this skill into the rules it gives an agent
16 rules. About 0 are common, 11 are distinctive, and 5 are very specific to this project.
≈0
Common · estimate
Many files like this have a rule of this kind.
≈11
Distinctive · estimate
This author's own choices. Worth a look.
5
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- Distinctive: line 28
When the user asks to "create a handover" or "save session state"
- Distinctive: line 29
When ending a work session on a multi-phase plan
- Distinctive: line 30
When work is interrupted and will be continued later
- Distinctive: line 31
When switching between team members or agents
- Distinctive: line 33
Do NOT generate handovers automatically.
- Distinctive: line 49
If no formal plan exists, create a standalone handover (see below)
- Very specific: line 78
If multiple handovers on the same day, append a counter: session-2025-02-14-2.md.
- Distinctive: line 92
After creating the handover:
- Very specific: line 94
Ensure todo.md reflects the actual current state
- Distinctive: line 96
Add any newly discovered items
- Very specific: line 101
Use the question tool to ask:
- Distinctive: line 107
If there is no formal plan but a handover is needed:
- Very specific: line 109
Create docs/handovers/session-<date>.md (in the docs directory instead of plans)
- Distinctive: line 115
Accuracy over completeness: Only document what actually happened.
- Distinctive: line 120
Use git for file state: Use git status and git diff --stat to identify modified files rather than relying on memory.
- Very specific: line 123
Todo updates stay lightweight: When updating todo.md in Step 4, make simple status changes directly.
≈ “Common” is an estimate: at least 1 in 5 of the 350,495 skill files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.
Files
Skill: Generate Handover
What This Skill Does
This documentation shall serve Agents and Humans when working in consecutive sessions with the project.
Creates a session handover document (plans/<name>/handovers/session-<date>.md) that captures everything a new session needs to continue work seamlessly:
- What was accomplished in this session
- What is in progress (interrupted work)
- Key decisions made and their rationale
- Current state of modified files
- Blockers and issues discovered
- Concrete next steps
- Context that would be lost without the handover
When to Use
- When the user asks to "create a handover" or "save session state"
- When ending a work session on a multi-phase plan
- When work is interrupted and will be continued later
- When switching between team members or agents
Do NOT generate handovers automatically. This skill is invoked manually by the user.
Execution Model (Recommended)
- Preferred: the primary agent runs this skill and writes the handover under
plans/<name>/handovers/. - Rationale: a handover is a distilled session narrative (decisions, rationale, state). It should be authored where the session context lives.
- Optional: delegate
git status/git diff --statinterpretation or bulk file listing todoc-explorer, but keep narrative/decisions in the primary.
Workflow
Step 1: Identify the Context
Determine what to capture:
- Which plan is being worked on? Check
plans/directory - What phase is active? Read
plan.mdandtodo.md - If no formal plan exists, create a standalone handover (see below)
Step 2: Gather Session Information
Collect from the current session context:
Progress:
- What tasks were completed? (from todo list changes, commits, conversation)
- What is currently in progress? (incomplete work, partial implementations)
- What was planned but not started? (remaining todo items)
Decisions:
- What technical decisions were made? (architecture, library choices, approach)
- What alternatives were considered? (and why they were rejected)
- What trade-offs were accepted?
Implementation State:
- Which files were modified? Use
git diff --statorgit status - Are there uncommitted changes?
- What tests are pending?
Issues:
- Are there blockers? (dependencies, unclear requirements, technical issues)
- Were workarounds applied that need revisiting?
Step 3: Write the Handover Document
Create plans/<name>/handovers/session-<YYYY-MM-DD>.md:
If multiple handovers on the same day, append a counter: session-2025-02-14-2.md.
Fill in all sections from the template:
- Session Summary: 2-3 sentences describing what this session was about
- Progress: Organized by completed / in-progress / not-started
- Key Decisions: Table with decision, alternatives considered, and rationale
- Current State: Modified files with what changed, pending tests
- Blockers & Issues: What prevents progress, with potential solutions if known
- Next Steps: Ordered list of what the next session should do first
- Context for Next Session: Free-form section for anything that would be lost - partial reasoning, important constraints discovered, "watch out for" notes
Step 4: Update the Todo List
After creating the handover:
- Ensure
todo.mdreflects the actual current state - Mark completed items
- Add any newly discovered items
- Note blockers
Step 5: Confirm with User
Use the question tool to ask:
- Is anything missing from the handover?
- Are the next steps correctly prioritized?
Standalone Handovers (No Plan)
If there is no formal plan but a handover is needed:
- Create
docs/handovers/session-<date>.md(in the docs directory instead of plans) - Omit plan/phase references from the frontmatter
- Focus on the general session context, decisions, and next steps
Rules
- Accuracy over completeness: Only document what actually happened. Don't invent or assume progress.
- Concrete next steps: "Continue working on auth" is useless. "Implement the JWT refresh token rotation in
src/auth/refresh.ts, starting from therotateTokenfunction stub" is actionable. - Decision rationale matters: Capture why decisions were made, not just what was decided. The next session needs the reasoning to avoid re-evaluating the same options.
- File-based interface: The handover document is the interface. All context goes into the file, not into chat messages.
- Don't duplicate the todo list: Reference
todo.mdfor task status. The handover captures session context (decisions, state, blockers), not a repeat of the task list. - Use git for file state: Use
git statusandgit diff --statto identify modified files rather than relying on memory. - Multiple handovers are normal: Each session can produce its own handover. They are additive, not replacements.
- Keep it scannable: The next session's agent will read this to bootstrap context. Use clear headings, short bullet points, and tables where appropriate.
- Todo updates stay lightweight: When updating
todo.mdin Step 4, make simple status changes directly. For complex plan restructuring, theupdate-planskill should be used instead.
Templates
This skill includes a normative template as a bundled file. Only read the templates when processing them. Output MUST follow the template headings and frontmatter keys:
tpl-session-handover.md- Structure for handover documents
Does it work?
Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.
No reviews yet
People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.
- It worked
- Partly
- It didn't
Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.
Your trail
Files you open will collect here, so you can find them again.
Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.
Now do this for your own.
Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.

