paseo-handoff
getpaseo/paseo
Hand off tasks to another agent with full context preservation.
What is paseo-handoff?
Transfers the current task—including context, decisions, and failed attempts—to a fresh agent. Use when the user requests a handoff or wants to delegate work to a specialized agent. The receiving agent gets a self-contained briefing so it can work independently.
- Parse user intent to select the receiving agent profile or create one via provider discovery
- Generate a structured handoff briefing covering task, context, relevant files, current state, and constraints
- Create an isolated workspace (worktree option available) for the delegated work
- Launch the receiving agent with zero context but complete task semantics
- Preserve investigation-only, fix, or refactor intent from the original request
How to install paseo-handoff
npx skills add https://github.com/getpaseo/paseo --skill paseo-handoff- Paseo skill installed and configured with agent profiles
- Access to `list_profiles` to review available receiving agents
- Current workspace context (files, decisions, state)
How to use paseo-handoff
- 1.Call `list_profiles` to review available agent profiles and their notes
- 2.Specify the receiving agent profile name, or let the skill choose based on task fit
- 3.Optionally request isolation with 'in a worktree' or 'worktree' in your handoff request
- 4.Provide a task description covering what needs to be done
- 5.The skill generates a structured briefing (task, context, files, state, attempts, decisions, criteria, constraints) and creates the receiving agent in a dedicated workspace
- 6.Review the new agent and workspace; it remains a subagent until you manually detach it
Use cases
- Delegate a partially-completed task to a specialist agent when you lack expertise
- Hand off a bug fix to a fresh agent after initial investigation reveals the root cause
- Transfer a refactoring task to another agent with explicit constraints on scope
- Pass a complex task to an agent in an isolated worktree to avoid conflicts
- Escalate a task when the current agent is blocked or needs a different skill set
- Coding agents managing multi-agent workflows
- Teams using Paseo for agent orchestration
- Users coordinating work across specialized agent profiles
paseo-handoff FAQ
The skill reviews configured profiles and their notes, then selects the best match for the work. If no profile fits, it uses Paseo's provider discovery fallback to create one.
Yes. Include 'in a worktree' or 'worktree' in your handoff request to create an isolated workspace with a short branch name derived from the task.
The receiving agent starts with zero context but receives a self-contained briefing that includes relevant files, current state, what was tried, and constraints. Include file paths and explanations in the briefing.
The new agent and workspace are returned to you. The agent remains a subagent in your track until you manually detach it; the skill does not wait or poll for completion.
The skill automatically carries investigation-only, fix, or refactor intent from your request into the handoff prompt, ensuring the receiving agent understands the exact scope.
Full instructions (SKILL.md)
Source of truth, from getpaseo/paseo.
name: paseo-handoff description: Hand off the current task to another agent with full context. Use when the user says "handoff", "hand off", "hand this to", or wants to pass work to another agent. user-invocable: true
Handoff Skill
Transfer the current task — context, decisions, failed attempts, constraints — to a fresh agent. The receiving agent starts with zero context, so the handoff prompt must be a self-contained briefing.
User's arguments: $ARGUMENTS
Prerequisites
Read the paseo skill. Call list_profiles before choosing the receiving agent. Do not create it until you have read the configured profiles and their notes.
Parsing arguments
- Agent profile — explicit profile name first; otherwise choose the profile whose
notesbest match the work. Materialize it intocreate_agentas described by the paseo skill. If no profile fits, use Paseo's provider discovery fallback. - Isolation — "in a worktree" / "worktree" → create a workspace with
isolation: "worktree", using a short branch name derived from the task. - Task description — anything else the user said.
The handoff prompt
The receiving agent has zero context. Include:
## Task
[Imperative description.]
## Context
[Why this task exists, required context.]
## Relevant files
- `path/to/file.ts` — [what it is and why it matters]
## Current state
[What's done, what works, what doesn't.]
## What was tried
- [Approach] — [why it failed or was abandoned]
## Decisions
- [Decision — rationale]
## Acceptance criteria
- [ ] [Criterion]
## Constraints
- [Must-not / must-preserve]
Preserve task semantics. Investigate-only → "DO NOT edit files." Fix → "implement the fix." Refactor → "refactor, not rewrite." Carry the user's exact intent.
Launch
Prepare the handoff in a dedicated workspace:
- Select the current workspace or call
create_workspacewith the requested isolation. - Call
create_agentwith a[Handoff] <task>title, the briefing as initial prompt, and the selectedworkspaceIdwhen explicit placement is needed. - Return the agent and workspace to the user, explaining that it remains in your subagent track until they detach it manually.
Do not encode independence as a create mode and do not invoke CLI or wire-level detach operations. Detach is a user gesture in the subagents track.
Do not wait or poll for the agent to finish.
Related skills
More from getpaseo/paseo and the wider catalog.

paseo-loop
Run an agent loop until an exit condition is met for iterative autonomous execution.

paseo-orchestrate
Deprecated. Renamed to paseo-epic. Loading this skill redirects to paseo-epic and tells the user.

release-beta
Cut a beta release of Paseo following the documented beta flow and checklist.

release-stable
Cut a stable release of Paseo—patch, minor, or promote from beta.

xcodebuildmcp
Agent skill from getsentry/mobilebuildmcp.

sentry-react-setup
Install and configure Sentry error monitoring, session replay, and tracing in React applications.