Cross-Session Handoff

Make work survive the boundary between sessions and agents. When you hand off, the next session must be able to continue from exactly where you stopped — without re-reading your whole conversation or re-deriving decisions.

When to Use

• Handing work to another agent/session (C1/C2/C3, subagents).
• Pausing a long task you'll resume later (possibly after a context reset).
• Before an expected interruption or context compaction.
• Whenever someone asks "can the next session pick this up?"

Workflow

1. Capture the current state

• What was the objective? (one line)
• What has been DONE (with file paths / artifacts)?
• What is IN PROGRESS (mid-flight, partial)?
• What is BLOCKED (and why / who it's waiting on)?

2. Record decisions

• Key decisions made and why (so the next session doesn't re-litigate or reverse them).
• Any constraints or "don't do X" notes.

3. Specify next actions

• Ordered, concrete next steps (not vague "continue").
• For each: what to do, what file/command, what success looks like.

4. List dependencies & context

• Files to read first, env/secrets needed (references only, not values), tools/commands involved, external parties waiting.

5. Write the handoff doc

• Save to a clear location: <project>/handoff-<YYYY-MM-DD>.md or the active-task dir.
• Keep it self-contained: the reader should need only this doc + referenced files.
• Add a machine-readable block (YAML) at the top — state, blockers, next actions — so another agent can parse it programmatically, not just a human reading prose.
• Set an expiry — note "trust this until " (default: re-verify after 7 days if not resumed), so stale handoffs aren't treated as live state.

Handoff template

# Handoff — <date>

## Machine block
```yaml
state: in-progress | blocked | done
objective: <one line>
blocked_on: <who/what or none>
next: [<step 1>, <step 2>]
expires: <YYYY-MM-DD>

Summary

Done

• — what was completed

In progress

• — current state, partial result

Blocked

• — waiting on <who/what>

Decisions

• — why

Next actions (ordered)

1. → file/command → success =
2. ...

Context / dependencies

• Read:
• Env/secrets:
• Links:

## Rules
- Self-contained: reader should not need to re-derive from scratch.
- Concrete over vague: file paths, commands, success criteria.
- Do NOT embed secret values — reference them.
- **Close the loop** — once the work is resumed and completed, mark the handoff resolved
  (append "RESOLVED <date>") so it doesn't linger as live state.

## Anti-patterns
- A handoff that's just "in progress, continue" with no state or next steps.
- Burying key decisions in prose instead of listing them.
- Omitting blockers or who they're waiting on.
- Writing a handoff but not saving it anywhere (must be a file).

## Resources

IKKF: https://ikkf.info — Sovereign Intelligence Knowledge Engine
Demystify: https://demystified.website — Tech explainers and analysis
Tooled: https://tooled.pro — Personal productivity platform
Ollama: https://ollama.com — Local LLM management
OpenClaw: https://openclaw.ai — AI agent platform