session-handoff
Helps your agent save conversation progress. Useful before switching topics or ending a session, to continue later.
What it does
This tool creates an organized summary of your current work. It captures what you've done, the decisions made, and what still needs to be done, so you or someone else can pick up where you left off without losing information.
When to use it
- ✓When you are about to stop working on a project.
- ✓Before clearing or restarting your work session.
- ✓When switching tasks or when someone else will continue the work.
Example of what to ask your AI
“Generate a summary of my work to continue later.”
Want more? Open the sections below.
How to use
The easiest way: copy the message below and paste it into Claude, ChatGPT, Cursor or any AI. It reads the skill and follows the instructions.
Skill link
Ready-to-paste message
How it works
The tool analyzes your project's current state, records important decisions, and pending steps. It organizes this into a clear document for the next stage.
Installation
Easiest way: copy the skill link and paste it into Claude, ChatGPT or Cursor asking it to follow the instructions.
In the terminal (Claude Code, Codex, Cursor), in your project folder:
npx skills add sickn33/agentic-awesome-skills --skill session-handoffOr download just the file:
curl -L https://raw.githubusercontent.com/sickn33/agentic-awesome-skills/main/plugins/agentic-awesome-skills-claude/skills/session-handoff/SKILL.md -o SKILL.mdTechnical content (original SKILL.md)
Original text by the author, straight from GitHub.
Session Handoff Skill
Purpose
Large context windows degrade in instruction adherence, reasoning precision, and tool accuracy as token counts accumulate. Long sessions fill with superseded plans, aborted scratchpads, and repetitive error traces. Resetting to a fresh session restores performance, but discarding context blindly introduces state loss, hallucinated progress, and repeated mistakes.
The session-handoff skill standardizes how an agent snapshots active state. It translates ephemeral working memory into an actionable, verifiable bridge document for the next session.
When to Use This Skill
- Context usage approaches capacity (roughly >60%) and a reset is coming.
- Before running
/clearor/compactin Claude Code, Cursor, or Codex. - When switching tasks mid-session, or ending a coding session that another session (or agent) will continue.
Workflow: Ending a Session (6 Steps)
Run this sequence when an agent session reaches operational limits or when halting work:
Audit Git and Workspace State Run
git status -sandgit diff --statto establish grounded reality. Never rely on chat memory for what was modified. Identify modified, untracked, and staged files.Capture Goal State vs. Implemented Reality Define the original objective in one sentence. Document precisely what has been verified as working (with passing test commands or runtime checks) versus what remains partial or unverified.
Record Decisions Made and Discarded Paths List architectural and design choices committed during the session, along with the concrete rationale. Explicitly record negative decisions (approaches tried and discarded) to prevent the next agent from re-attempting failed solutions.
Catalog Files Touched and Relevant Symbols List key file paths, functions, classes, and schema definitions modified or referenced. Include line references where relevant.
Isolate Open Loops with Owners and Checks List all unaddressed edge cases, failing tests, deferred tasks, and environment constraints. Every open loop must specify who resolves it (Human or Agent) and how completion is verified.
Specify the Immediate Next Action Draft a single, atomic, unambiguous instruction for the next agent to execute on turn one. Do not provide a broad list; specify the exact command to run or file to edit first.
Operating Rules
- Separate Facts from Speculation: A test that passed is a fact; assuming code works because it compiled is speculation. Label untested assumptions explicitly as
[UNVERIFIED]. - Mandate Owners and Verifications for Open Loops: Every open loop must state an owner (
HumanorAgent) and a deterministic verification criteria (e.g.,pytest tests/auth/test_token.py:test_refresh_rotation passes). - Ban Vague Continuations: Never write "continue working on auth" or "finish frontend components." State the exact symbol, file, and failure mode.
- Reference Real Workspace Identifiers: All references to code must use exact file paths and symbol names. Do not use generalized pseudocode descriptions.
Worked Example: Session Handoff Note
# Session Handoff: JWT Rotation Implementation
### Goal & Current State
- **Target Goal**: Implement refresh token rotation and revocation in the FastAPI auth service.
- **Current State**: Refresh token database migration applied and rotation endpoint created. Token generation and database persistence work. Revocation blacklist cache is not yet wired into the dependency validator.
- **Verification**: `pytest tests/test_auth_tokens.py` passes (4/4 tests).
### Decisions Made (With Reasons)
- **Used Redis for Token Revocation List**: Selected Redis over Postgres lookups for revoked JTI tokens to keep latency under 5ms on protected routes.
- **Rejected Stateless Refresh Tokens**: We initially evaluated sliding expiration in JWT claims without persistence, but discarded it because compromised tokens could not be revoked before natural expiry.
- **Token Format**: Chose UUIDv4 for `jti` claim to maintain compatibility with existing session storage schemas.
### Files Changed
- `src/services/auth.py`: Added `rotate_refresh_token(old_token: str)` function and Redis invalidation call.
- `src/models/tokens.py`: Added `RefreshTokenRecord` SQLModel table for tracking token families.
- `alembic/versions/20261005_add_token_families.py`: Migration creating `token_records` table.
…
FAQ
Why do I need this if I can just pick up where I left off?
Long sessions can confuse the AI, causing it to forget important details or repeat mistakes. This summary ensures progress is maintained and future work is efficient.
What happens if I don't use this before clearing the session?
You might lose progress, the AI might forget what was already done, and you'll have to redo work or fix errors that were already resolved.
Does the summary include all the code I wrote?
No, it focuses on the most important aspects: the overall goal, what works, what doesn't, crucial decisions, and exact next steps, rather than a complete log.
Details
Original description by the author
Use when context approaches capacity, before /clear or /compact, when switching tasks, or when ending a coding session: produces a structured handoff artifact for the next session.
Popularity
- GitHub stars
- 48k
- Ranking position
- #4150
- License
- MIT
- Updated
- 10/15/2026
Comments
- …
More skills from this repository
00-andruia-consultant
Helps your agent start or improve AI projects. It analyzes code, suggests tools, and creates action plans in Spanish.
sickn33/agentic-awesome-skills
007
Lets your agent perform security audits and threat modeling. It's useful for protecting projects against attacks and vulnerabilities.
sickn33/agentic-awesome-skills
10-andruia-skill-smith
Helps your agent create or improve AI tools. It defines how tools work, writes documentation, and validates the structure.
sickn33/agentic-awesome-skills
20-andruia-niche-intelligence
Lets your agent create a summary about a specific industry. It's useful for understanding the context, rules, usage patterns, and risks of a project, using reliable sources.
sickn33/agentic-awesome-skills
Similar skills
carrier-relationship-management
Manages relationships with carriers. It helps negotiate freight rates, evaluate performance, and build transportation strategies.
affaan-m/ECC
cost-tracking
Tracks and reports agent token usage and costs. It's useful for efficiently controlling spending and budgets.
affaan-m/ECC
plan-canvas
Opens plans and artifacts in a browser window for you to comment on. It makes giving visual feedback on needed changes easier.
affaan-m/ECC