Skip to main content
skill CX Codex

plan-arbiter — Claude Code skill: what it does, how to install

plan-arbiter compares competing implementation plans from agents like Codex and Claude Code, merging the strongest ideas into a single roadmap.

MIT 297 installs ★ 4,679 updated 2026-10-22

Install into Claude Code · Skills CLI (Vercel) All methods and platforms ↓
npx skills add BuilderIO/skills --skill plan-arbiter -a claude-code
Open on GitHub

What plan-arbiter does

The skill ingests multiple proposals from pasted text, local files, session IDs, PR descriptions, or chat transcripts. It extracts core objectives, assumptions, targeted files, and validation steps so disparate plans become directly comparable.

Next, it cross-reviews each proposal against the actual codebase, docs, and existing tests. It avoids favoring length over substance. Decisions result in adopting one option, building a hybrid path, or sending both back for revision when core requirements are missing.

Planning remains strictly read-only until you instruct the agent to execute. Without explicit permission to apply changes, the skill stops after delivering the structured decision memo.

What it helps with

  • resolve conflicting architecture proposals from multiple AI models
  • combine the strong system design of one plan with another's rollback strategy
  • cross-reference theoretical steps against real repo files and tests
  • generate a clear decision memo specifying why rejected ideas were discarded

Example prompts for the agent

  • Compare the refactoring plans in plan-a.md and plan-b.md and create a unified execution strategy.
  • Review the proposed database migration from Codex against Claude Code's session transcript and pick a winner.
  • Cross-review these two pasted approaches and highlight hidden risks in our codebase.

How to install plan-arbiter

1Skills CLI (Vercel)

npx skills add BuilderIO/skills --skill plan-arbiter -a claude-code

2Manually: put the files in place

~/.claude/skills/plan-arbiter/SKILL.md

Copy the whole skill folder; for one project use .claude/skills/plan-arbiter/

Step by step, with commands for each platform: How to install skills in Claude Code: five ways

Questions about plan-arbiter

How do I install plan-arbiter in Claude Code?

Run npx skills add BuilderIO/skills --skill plan-arbiter -a claude-code. The skill will be installed at ~/.claude/skills/plan-arbiter/SKILL.md.

What does plan-arbiter do?

It evaluates competing agent proposals against your project files and produces a single actionable plan with rejected alternatives listed.

Does plan-arbiter automatically modify project files?

No, planning runs in read-only mode by default. The agent will only touch project code if you explicitly ask it to proceed with implementation.

Source text

SKILL.md · MIT
---
name: plan-arbiter
description: Use when asked to compare, cross-review, merge, judge, choose, or arbitrate competing plans from multiple agents such as Codex and Claude Code; when given two or more proposed plans, session IDs, transcripts, plan documents, PR descriptions, or pasted strategies; or when the user wants one recommended execution plan after agents review each other's proposals.
---

# Plan Arbiter

Turn competing plans into one executable direction. Preserve the best ideas,
reject weak assumptions, and produce a clear handoff instead of a blended mush.

## Workflow

1. Collect the source plans.
2. Normalize each plan into comparable claims.
3. Cross-review the plans against each other and the real codebase or task
   context.
4. Choose a winner, merge a better hybrid, or send the plans back for revision.
5. Produce one execution handoff with verification gates and rejected
   alternatives.

Planning is read-only unless the user explicitly asks you to implement after the
decision.

## Collect Source Plans

Accept plans as pasted text, local files, session IDs, transcript paths, PRs,
comments, visual-plan links, or chat history. Resolve the original artifacts
when possible so you can see prompt changes and assumptions that may be missing
from a final summary.

If a plan is still being written and the user asked you to wait, monitor it
until it is done or blocked. If a plan cannot be resolved, continue with the
available plan text and mark the missing source as a risk.

## Normalize

For each plan, extract:

- Objective and scope.
- Key assumptions and unresolved questions.
- Proposed files, modules, APIs, data shapes, UI states, or workflows.
- Implementation sequence.
- Validation strategy.
- Rollback or migration concerns.
- Cost, complexity, and expected executor fit.

Do not reward verbosity. Prefer plans that are concrete, grounded in real code,
and honest about tradeoffs.

## Cross-Review

Review each plan as if another capable agent wrote it:

- Check whether it satisfies the user's actual request.
- Verify claims against the repo, docs, tests, screenshots, or external systems
  when those are relevant and available.
- Identify hidden dependencies, missing tests, risky sequencing, vague steps,
  unnecessary scope, and hard-to-reverse decisions.
- Notice complementary strengths: one plan may have the better architecture
  while another has the better migration or validation path.
- Separate plan quality from executor preference. A cheaper/faster executor can
  be the right choice for implementation even when another model produced the
  best critique.

Use subagents for independent review when the plans are large, the codebase is
wide, or the decision would benefit from separate technical and product passes.

## Decide

Choose one of three outcomes:

- **Adopt:** pick one plan mostly as written.
- **Hybrid:** combine specific pieces into a stronger execution plan.
- **Revise first:** request another planning pass because both plans miss a
  key constraint or depend on an unresolved decision.

Use this tie-break order:

1. Correctness and fit to the user's request.
2. Grounding in real files, APIs, tests, data, and UI behavior.
3. Simpler first implementation that does not block the intended future.
4. Better validation and rollback story.
5. Lower token/time cost for execution once quality is acceptable.

## Handoff

Return a compact decision memo:

```md
Decision
- Adopt Plan A / Hybrid / Revise first.

Why
- The deciding evidence and tradeoffs.

Execution Plan
- Ordered steps with files or surfaces to touch.

Borrowed From Other Plans
- Useful pieces kept from non-winning plans.

Rejected
- Ideas intentionally not taking, with reasons.

Verification
- Tests, browser checks, screenshots, CI, review, or deploy checks needed.

Executor Recommendation
- Which agent/model should implement and why.
```

When the user already asked for execution and the chosen path is clear, proceed
with the selected plan after reporting the decision briefly. Otherwise stop at
the handoff and ask for approval.

© BuilderIO · MIT · BuilderIO/skills

Similar skills

All skills for Agent setup: memory, context, orchestration →

From the same set11

The whole set →
efficient-fable efficient-fable is a Claude Code skill that delegates to efficient-orchestrator only on explicit command invocation. CX Codex +1 900installs quick-recap quick-recap adds a short final status block to completed agent responses. Use it when project instructions require a consistent completion signal. CX Codex +1 698installs read-the-damn-docs read-the-damn-docs makes the agent search for and read current official documentation before writing code instead of trusting model memory. CX Codex +1 567installs stay-within-limits stay-within-limits — a skill for Claude Code, Codex, Cursor and other platforms that enforces 5-hour and weekly usage limits during long-running or parallel agent work. CX Codex +1 350installs agent-watchdog agent-watchdog — a skill to review and improve another agent's output. Give it a session ID, PR, or logs—it investigates the same problem independently, compares results against the original solution, and flags gaps. CX Codex +1 331installs plow-ahead plow-ahead — a skill that lets your agent keep going through ordinary ambiguity without asking at every step. Use it when you want to hand off a task and return to a finished result. CX Codex +1 250installs visual-edit visual-edit is a Claude Code skill that opens a running local app in Design for visual inspection and collaborative edits. It activates when someone asks to view, share, or change a real app in the editor. CX Codex +1 159installs rewind The rewind skill for Claude Code retrieves local screen context using Clips Desktop. The agent runs it when asked about recent events or previous screen activity. CX Codex +1 123installs

Project sponsors

Companies that keep this analytics open