# Agent Skill Bundle — Agent Handoff

> Generated from the canonical public Skill source. This bundle is an invocation artifact, not a separate editable source.

## Agent instruction

Apply this Skill to the user’s actual task. Follow its inputs, behavior, guardrails, outputs, and verification criteria. Do not merely explain or summarize the Skill. If required task context is missing, ask only for the missing information that blocks correct execution.

## Skill identity

- source: skills/agent-handoff
- canonical human page: https://wbeen-personal-kb.vercel.app/skills/agent-handoff/
- version: 0.1.0
- workflow phase: execute
- invocation: user

## Skill contract

Transfer an in-progress AI-assisted task to another agent or session without losing decisions, evidence, boundaries, and next actions.

### Inputs

- Current objective and scope
- Decisions already made
- Completed and incomplete work
- Evidence, identifiers, URLs, or artifacts needed to resume
- Known risks and boundaries

### Outputs

- Compact handoff brief
- Verified current state
- Explicit unresolved items
- Next executable action

### Guardrails

- Preserve exact identifiers when they are required for continuation.
- Separate verified state from assumptions or recommendations.
- Do not include secrets, credentials, or unnecessary private source content.
- Do not silently rewrite an unresolved decision as if it were settled.

### Verification

- A new agent can identify what is done and what is not done without asking for a full recap.
- The next action can be executed directly from the handoff.
- Important evidence and constraints are traceable to their source or artifact.

### Source instructions

## What it does

Agent Handoff preserves **operational continuity**, not conversational history. It captures the minimum state another agent needs to resume the same task correctly: objective, decisions, evidence, boundaries, unfinished work, and the next executable action.

## When to reach for it

Use it before switching tools, computers, sessions, or agents, especially when the work already contains decisions that would be costly or risky to rediscover.

## What one run does

1. Restate the objective and current scope.
2. Separate completed, in-progress, and unresolved work.
3. Preserve decisions and exact identifiers that materially affect continuation.
4. Attach or point to the evidence needed to trust the current state.
5. State the next executable action and any stop conditions.

## Common failure modes

- Writing a narrative recap instead of an operational state packet.
- Omitting failed attempts or unresolved constraints.
- Copying excessive private context that the next agent does not need.
- Leaving the receiving agent with several possible next actions instead of one clear continuation point.

## It’s working if

The receiving agent can resume the same task with minimal rediscovery and can distinguish confirmed state from assumptions.

## Where it fits

This is primarily an **Execute**-phase continuity skill. Once resumed work reaches a completion boundary, Evidence-first Review is a natural next step.

## Completion rule

Return the useful task result first. Then state briefly which verification criteria passed, which could not be checked, and any stop/escalation condition that remains.