planx

A command that writes a concise implementation plan into a dated directory under docs/plans. The plan is intended for another AI to follow without needing the surrounding conversation.

In plain words
What is it for?
Use it when you want to plan a feature or fix in the ui-debugger-mcp project without editing code or running implementation commands.
Why use it?
It records the task, relevant files, architecture choices, tests, and documentation work before implementation begins.

Command for Claude Code

Install

Getting it into your agent

One page per mod, every tool's command on it. A separate URL per tool would split the same page into five that compete with each other.

agentmods
npx agentmods add commands/developerz-ai/ui-debugger-mcp/planx
Clone the repo
git clone --depth 1 https://github.com/developerz-ai/ui-debugger-mcp

Made for: Claude Code.

Per session 30 Only the description is in the session, so the agent can decide to use it. The body loads when it is invoked.
When invoked 1,784 The whole file, excluding the scripts and references it only reads on demand.
Security scan A 0 findings. Scan, not verified.
Origin original No closer match found in the catalogue.
Token cost

What it costs to keep this loaded

Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.

ModelPer sessionOnce invoked
Fable 5 $0.00030 $0.01784
Opus 5 $0.00015 $0.00892
Sonnet 5 $0.00006 $0.00357
Haiku 4.5 $0.00003 $0.00178

Measured 2d ago against content hash a0481ae1bbfc, method: parsed. Prices are Anthropic first-party input rates as of 2026-08-30, from the pricing page.

Security

Grade A, and why

planx scanned grade A with 0 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 2d ago.

A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.

Nothing flagged

None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.

.claude/commands/planx.md · 105 lines

How it starts

The opening of the file, as written. The whole thing — 105 lines — stays where its author put it; the contents beside it link to each section on GitHub.

/planx

Produce a concise plan another AI can execute with zero extra context. Plan only — no implementation, no code execution, no edits outside the plan dir.

Goal

$ARGUMENTS

Steps

  1. Resolve path. Run date +%Y, date +%m, date +%d. Dir = docs/plans/<YYYY>/<MM>/<DD>/. Glob docs/plans/<YYYY>/<MM>/<DD>/1* → next number = highest existing 1NN-* + 1, else 101. Slug = kebab-case title, max 5 words. Final plan dir: docs/plans/<YYYY>/<MM>/<DD>/<1NN>-<slug>/.

  2. Explore. Task (subagent_type=Explore, thoroughness="very thorough"): existing patterns + files to touch (file:line), the right layer under src/* (mcp tool defs, agent loop + prompts/, adapters/{browser,desktop,android}, session, config, services, findings, cli), tests (unit *.test.ts vs e2e e2e.test.ts / dummy-web.e2e.test.ts), the shared adapter contract, Zod schemas at each boundary, custom error classes in src/errors.ts, and the docs that move with the code (docs/idea/*, docs/reference.md, README.md, CLAUDE.md). Prefer codegraph_* for structural lookups when a .codegraph/ exists. Skip only for trivial asks.

  3. Write the plan as multiple files in the plan dir — never one big plan.md. Always produce an overview.md index plus one <NN>-<aspect>.md per separable area (e.g. 01-adapter-contract.md, 02-agent-tools.md, 03-prompts.md, 04-mcp-surface.md, 05-config.md, 06-tests.md). Split by area of work so each file is independently executable and stays short. Match the existing house style in docs/idea/ — terse fragments, file:line refs, tables.

    overview.md — the map. Sections:

# <Title>

## Goal
1-2 sentences: what + why.

## Context
- Stack facts the executor needs (Bun + TS, MCP stdio server, Vercel AI SDK agent loop running *inside* the server, OpenAI-compatible router / OpenRouter default, CDP for web / X11-Wayland for desktop / ADB for android behind one adapter contract, Zod at every boundary — only what's relevant).
- Which actor it touches: smart agent (caller) / fast guy (driver) / vision guy (eyes) — see `docs/idea/models.md`.
- Reference patterns: `src/<layer>/<thing>.ts:12` — follow this for Z.

## Plan files (execute in order)
1. [`01-<aspect>.md`](01-<aspect>.md) — one line: what it covers.
2. [`02-<aspect>.md`](02-<aspect>.md) — ...

## Done when
- Verifiable acceptance criteria spanning the whole feature.

## Risks / open questions
- Anything the executor must decide or watch.

Read the full file on GitHub · 105 lines

Changes

What this file has done since we first saw it

Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.

  1. 2d ago First seen · 105 lines · 30 tokens per session scan A a0481ae1bbfc

Subscribe to this mod's changes

planx is a command published in the GitHub repository developerz-ai/ui-debugger-mcp (2 stars, last pushed 25d ago), licensed MIT. It adds 30 tokens to every session and 1,784 once invoked, about $0.0002 per session on Opus 5. A static security scan graded it A with 0 findings. No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-31.