- Home
- Skills
- Agents & Orchestration
- Durable Session Handoff
Durable Session Handoff
Moves a long coding session's project knowledge into three durable files.If it saves you a session, a star on the repo is very much appreciated — thank you.
Free
Works with the AI Agents you already use
Durable Session Handoff
Example session with this skill installed
The login feature is done and the tests pass. I want to start a fresh session
for the dashboard work — hand this one off first.
- Read your context and instructions
- Compiled the durable session handoff
Re-ran verification before writing anything (after a long session, a memory of
"tests passed" is not evidence), then wrote the three handoff files.
Verification, re-run just now
| Check | Result |
|---|---|
npm run build | exit 0, built in 4.2s |
npm test src/auth/__tests__/login.test.ts | 14 passed, 0 failed |
git status --short | 2 modified, 0 untracked |
AGENTS.md — idempotent block, appended
<!-- SESSION_HANDOFF:START -->
## Read this first
1. Read `docs/STATUS.md`, then restate: what this project is, its status, the next step.
2. Read `docs/DECISIONS.md` before changing auth behaviour.
3. If you can't restate something, that's a documentation gap — ask, don't guess.
## Constraints
- Session tokens live in httpOnly cookies; never put them in localStorage.
- The refresh endpoint must stay idempotent (D-004).
- `src/auth/legacy/` is read-only until the migration lands.
## Commands
- build: `npm run build`
- test: `npm test`
- verify login flow: `npm test src/auth/__tests__/login.test.ts`
## Known traps
- Mocking `fetch` breaks the token-refresh test; stub the client instead.
<!-- SESSION_HANDOFF:END -->
Connects securely to your tools. The creator never sees your data.
About this skill
Moves a long coding session's project knowledge into three durable files (AGENTS.md, docs/STATUS.md, docs/DECISIONS.md) so a fresh session can pick up where you left off without the old conversation. Re-runs verification instead of trusting memory, and includes a cold-read check that proves the handoff actually worked. Works with any agent that supports the open SKILL.md format.
Free and MIT licensed. Source, full docs and an honest validation status: https://github.com/fanpaisk/session-handoff
If it saves you a session, a star on the repo is very much appreciated — thank you.
Requires local file system access. Cannot automatically rotate secrets found in history. Effectiveness depends on the agent's ability to read AGENTS.md in the next session.
How to use
- 1
Connect your agent to the Agensi MCP
Copy the prompt and paste it into Claude, ChatGPT, Cursor or any agent that supports MCP. It sets up the connection for you. Manual setup
- 2
Sign in to Agensi
Your agent shows a sign-in link. Sign in and click Allow.
- 3
Ask for what you need
Your agent loads this skill and uses it on your task. It always asks before any payment.
Reviews
No reviews yet
Be one of the first to try it. Every listed skill goes through the trust checks below.
Security scanned
Checked by our 8-point scan before listing
Fresh listing
Recently published to Agensi
Free forever
No account required to browse
Trust & safety
Security scan
Security scan passed 1 day ago
- Free with an Agensi account
Requires
Permissions
File Scopes
Terminal/Shell: the skill re-runs your project's build, test and acceptance commands at handoff time instead of trusting the session's memory of them — that verification step is the core of what it does. Write Files: it creates AGENTS.md plus docs/STATUS.md and docs/DECISIONS.md, and may add an authority pointer to an older progress document. Read Files: it must read the existing instruction and status files before rewriting them, so it never destroys content you wrote yourself. Network Access / Browser / Environment Variables: not used at all. The skill has no runtime dependencies and makes no outbound requests.