Skip to content

Handoff and Wayfinder: Useful Agent Skills Without Adopting an Entire Methodology

Research date: Aug 21, 2026. This article compares the named skills in Matt Pocock's Skills with Superpowers, using their primary documentation. The recommendations are editorial synthesis.

TL;DR

Handoff is a small, user-invoked skill that writes a portable document so a fresh agent, another harness, or a human can resume work without reconstructing the conversation. It is useful because it creates one clear boundary artifact and points to durable work instead of copying it.

Wayfinder is a deliberately heavier, user-invoked planning skill for a destination that is too uncertain or too large for one agent session. It keeps a shared issue-tracker map of decisions, resolves one frontier decision at a time, and hands the resulting plan to an implementation flow.

Superpowers is a separate project and a complete development methodology. Its automatic, mandatory flow can add brainstorming, worktrees, detailed plans, subagents, TDD, review, and branch-finalization to ordinary work. That rigor may be valuable for autonomous, high-risk changes, but it is more process than most tasks need.

The practical default: adopt Handoff first; use Wayfinder only for large, foggy work; borrow individual Superpowers practices rather than installing its entire governing workflow.


The three things are not the same

Tool Job Scope Invocation model
Handoff Move working context safely A session, agent, directory, branch, or harness boundary Explicitly user-invoked
Wayfinder Find a route through an ambiguous multi-session effort Planning before implementation Explicitly user-invoked
Superpowers Govern an end-to-end coding workflow From idea through implementation and branch completion Skills are intended to trigger automatically

Handoff and Wayfinder are skills from the same library. Superpowers is a distinct project; they share the broader agent-skills ecosystem, not a common workflow or dependency.

Handoff: a narrow context-transfer primitive

The handoff skill compacts the current conversation into a Markdown handoff document for a fresh agent. Its purpose is portability, not compression.

The document should:

  • state what the next session should do;
  • name suggested skills for that session;
  • reference existing specs, plans, issues, commits, and diffs rather than duplicate them;
  • redact sensitive information; and
  • be written to the operating system's temporary directory, not the project workspace.

This is useful whenever conversational context stops being the correct source of truth: a session reaches its limit, an investigation branches, work moves to another agent, or a human needs to take over. Durable project artifacts remain where they belong; the handoff is the compact pointer layer that explains how they fit together.

When to use it

Use Handoff after a meaningful boundary, not as routine note-taking:

  • before asking a fresh agent to continue a partially completed task;
  • when transferring work between coding-agent harnesses;
  • before delegating a focused side investigation;
  • when pausing an effort that already has durable artifacts; or
  • when someone else needs the intent, current state, and next question.

For a small, self-contained edit in one session, it is unnecessary overhead.

Wayfinder: map decisions, then stop

The wayfinder skill addresses a different failure mode: the destination is known, but the route cannot be responsibly planned in one session.

It creates a map issue with child decision tickets. A ticket answers a question that changes the route; it is not a pre-cut implementation task. The map tracks:

  • the destination and standing notes;
  • decisions already made;
  • the frontier: open, unblocked, unclaimed decision tickets;
  • the fog of war: in-scope questions that are not yet precise enough to ticket; and
  • consciously excluded work.

Tickets may be research, prototype, grilling, or unblock-a-decision tasks. Each session takes one decision from the frontier, records and closes it, then promotes newly clear questions from the fog into tickets. Wayfinder is planning by default: once the route is clear, it hands off to specification and implementation rather than continuing into code.

When Wayfinder earns its cost

Use it when all of these are true:

  1. The effort will exceed one agent session.
  2. Important decisions are genuinely unknown, not merely unimplemented.
  3. The decisions can be shared and tracked on an issue tracker.
  4. Resolving one decision will reveal what the next decision should be.

Do not use it for a normal feature with a known design. A short spec, a small plan, or direct implementation is better. The skill itself describes Wayfinder as its heaviest flow; its value comes from avoiding false certainty, not from making every task look large.

Why this can feel better than Superpowers

Superpowers calls itself a complete software-development methodology. Its documented baseline workflow is:

  1. brainstorm and approve a design;
  2. create an isolated worktree;
  3. write a detailed implementation plan;
  4. execute via subagents or batches;
  5. apply red-green-refactor TDD;
  6. request review; and
  7. verify and finish the development branch.

Its README says the agent checks for relevant skills before every task and describes these as mandatory workflows rather than suggestions. That policy makes the framework coherent, repeatable, and attractive for long autonomous changes. It also means even a straightforward request can acquire design gates, worktree setup, planning artifacts, subagent coordination, and test-first constraints.

The difference is chiefly default scope and control:

  • Handoff adds one artifact only when context must move.
  • Wayfinder adds a map only when uncertainty outgrows a session.
  • Superpowers establishes an operating system for most coding work.

So “Superpowers is over-engineered” is not a claim that its practices are bad. It means its baseline governance is larger than the problem boundary for many tasks.

A pragmatic adoption policy

Situation Best fit
A task spans sessions, people, or harnesses Handoff
A known change needs implementation Use the host agent's normal planning and verification flow
A large initiative has a destination but unclear decisions Wayfinder
A risky, long-running autonomous change needs strong discipline Selectively adopt Superpowers practices, or use Superpowers if the team accepts its full workflow

Start with Handoff because it has the smallest behavioral footprint and protects against a common agent failure: losing the reason behind existing artifacts. Add Wayfinder when the work has real fog of war. Treat Superpowers as an intentional team-process choice, not a dependency required to gain good planning, tests, or review.

Sources