agentmods.dev

Planchette: Instructions file for Claude Code

AGENTS.md

Planchette AGENTS.md is an instructions file for Claude Code, Codex, OpenCode from marcello-a/Planchette. It costs 7,940 tokens per session, scanned B, original, MIT.

Contributor instructions for Planchette, a native macOS terminal IDE for running multiple AI coding-agent sessions and organizing them by project.

In plain words
What is it for?
Use them when changing Planchette's macOS app, terminal layouts, session status, updater, localization, or AI summaries.
Why use it?
They document important boundaries, such as avoiding new dependencies, keeping the pinned Ghostty engine, and localizing user-facing text.

Instructions file for Claude CodeCodexOpenCode

Written for Claude Code and Codex and OpenCode: SessionEnd hook event, but also the file is AGENTS.md. Also seen: reads .claude/ paths; mentions Claude Code; mentions AGENTS.md.

This is marcello-a/Planchette's own configuration. It tells Claude Code, Codex and OpenCode how to work on Planchette itself, so it is not a mod to install elsewhere. Copy it as a starting point and replace the rules that are about this project. Everything Planchette configures →

not rated 7repo changed today B scan 7,940 tokens original MIT
View source ↗ marcello-a/Planchette
Reuse

Borrowing it

Nothing to install: this file belongs to marcello-a/Planchette. Take a copy, put it at the same path in your own repository, and replace the rules that are about this project with yours.

Copy the file
curl -O https://raw.githubusercontent.com/marcello-a/Planchette/main/AGENTS.md
Clone the repo
git clone --depth 1 https://github.com/marcello-a/Planchette

Made for: Claude Code, Codex, OpenCode.

Wrote this? Show the measurements

A badge with what this costs and how it scanned, read live from this page, so it follows the numbers instead of freezing them. Markdown for a README, HTML for a documentation site or a project page.

agentmods badge for Planchette AGENTS.md

README.md
[![agentmods](https://agentmods.dev/badge/instructions/marcello-a/planchette/agents-md/github.svg)](https://agentmods.dev/instructions/marcello-a/planchette/agents-md)
Your own site
<a href="https://agentmods.dev/instructions/marcello-a/planchette/agents-md"><img src="https://agentmods.dev/badge/instructions/marcello-a/planchette/agents-md/github.svg" alt="Measured on agentmods" height="20"></a>

Or the 80×15 button, for a site that already has a row of RSS and ATOM ones. Only the verdict fits; the numbers stay here.

agentmods 80×15 button for Planchette AGENTS.md

Your own site · 80×15
<a href="https://agentmods.dev/instructions/marcello-a/planchette/agents-md"><img src="https://agentmods.dev/badge/instructions/marcello-a/planchette/agents-md.svg" alt="Reviewed on agentmods" width="80" height="20"></a>
Per session 7,940 This file is loaded in full into every session.
When invoked 7,940 The same file — it is already loaded in full.
Security scan B 1 finding. A grade says what 26 rules found in the file — not that it is safe.
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 5.1 $0.07940 $0.07940
Opus 5.5 $0.03176 $0.03176
Sonnet 5.5 $0.01588 $0.01588
Haiku 4.5 $0.00794 $0.00794

Measured today against content hash 07ca97e6b18e, method: parsed. Prices are Anthropic first-party input rates as of 2026-10-07, from the pricing page.

Security

Grade B, and why

Planchette AGENTS.md scanned grade B with 1 finding 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 today.

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.

Reads agent configuration directoriesmediumAgent snooping

.claude/, .codex/, .gemini/ hold keys, settings and other credentials a mod has no legitimate need for.

installer either: it would otherwise wire `~/.claude/settings.json` from a test
AGENTS.md · 384 lines

How it starts

The opening of the file, as written. The whole thing — 384 lines — stays where its author put it; the contents beside it link to each section on GitHub.

AGENTS.md — guide for AI contributors

This file orients an AI (or a new engineer) working in Planchette. Read it before making changes. It complements docs/ARCHITECTURE.md (how the code is structured) and CONTRIBUTING.md (how to build, test, release).

What Planchette is

A native macOS terminal IDE for running many AI coding agents at once, built on the Ghostty engine (libghostty). It shows which session needs attention (asking / done / free), keeps layouts persistent across restart & reboot, and groups terminals by project. See README.md and docs/CONCEPT.md.

Ground rules

  1. No new external dependencies without asking. The app deliberately links only Apple frameworks + our self-built GhosttyKit. The updater uses the GitHub REST API directly instead of Sparkle; AI summaries shell out to the user's existing claude CLI instead of an SDK. Keep it that way unless the user explicitly approves a dependency.
  2. libghostty is pinned. vendor/ghostty is a submodule fixed to a tag (currently v1.3.1). Its embedding API is not stable across versions — do not bump it casually; a bump is its own reviewed change.
  3. Everything user-facing is localized. Never hardcode a display string. Add an LKey case and a value in all language tables in Localization.swift (EN is the required fallback — a test enforces it).
  4. State changes go through AppState and must be persisted. If you add a persisted field, update PersistedState (both inits), saveNow, applyRestore, startFresh, and AppState.init's early load.
  5. A user-facing feature is not done until it is in the Help tab. Every feature the user can reach appears in Help.sections (Help.swift), which is what the searchable Help tab in Settings renders. Add the entry in the same change as the feature, reusing the LKeys you already added for the control itself — the catalogue deliberately holds keys, not its own copy of the text, so help cannot drift from the UI. A test walks every entry and fails on an untranslated key. If a feature has no tooltip worth listing, that is a sign the control needs one, not that the feature skips the catalogue.
  6. Keep the main thread free. Subprocess/network work (claude -p, AppleScript, lsof, GitHub API) runs off-main; only touch @MainActor state after awaiting back.
  7. Verify by driving the app, not just building. See "Verifying" below.
  8. A big new feature gets a hint — ask the user. When you finish a big user-facing feature (a new panel, a new way to work, not a tweak), ask once whether it should get a first-use hint, and offer to build it. A hint is one FeatureHint case plus its content in FeatureHintContent.of (Views/FeatureHintView.swift), a looping hints/<name>.gif that shows the benefit (recorded from the real app by a scripts/perf scenario, like notepad.py gif), one title and one line of body text in all 7 languages, a .featureHint(.x) on the control it points at, an offerHint(.x) at the moment the user does something that makes it useful (never from restore code), and completeHint(.x) where the feature is used. Text rule: short, concrete, makes you want to try it — understanding comes first.

Read the full file on GitHub · 384 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. today Changed · +1 lines · +35 tokens per session 07ca97e6b18e
  2. 7d ago Changed · +3 lines · +343 tokens per session 1e4dc643880f
  3. 14d ago Changed · +5 lines · +90 tokens per session 8e430b53f149
  4. 19d ago Changed · +19 lines · +547 tokens per session 245a0e85cb86
  5. 21d ago Changed · +29 lines · +785 tokens per session 9a8bae3fd96e
  6. 22d ago Changed · +75 lines · +1,861 tokens per session 5edcb36d67cc
  7. 23d ago Changed · +86 lines · +1,629 tokens per session scan A → B 157423510ec2
  8. 2mo ago First seen · 166 lines · 2,650 tokens per session scan A 0256d9d540cf

Subscribe to this mod's changes

Planchette AGENTS.md is an instructions file published in the GitHub repository marcello-a/Planchette (7 stars, last pushed yesterday), licensed MIT. It adds 7,940 tokens to every session, about $0.0318 per session on Opus 5.5. A static security scan graded it B with 1 finding (reads agent configuration directories). No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-31.

Building agents
Related

Planchette AGENTS.md compared with instructions from other repositories

Planchette AGENTS.md next to 6 instructions from other repositories
ModTypeStarsLast pushLicenceScanTokensWorks with
Planchette AGENTS.mdmarcello-a/PlanchetteInstructions file7yesterdayMITB7,940Claude Code, Codex, OpenCode
vscode buildNext.instructions.mdmicrosoft/vscodeInstructions file195ktodayMITA9,719GitHub Copilot
next.js AGENTS.mdvercel/next.jsInstructions file143kyesterdayMITA7,461Codex, OpenCode
langchain AGENTS.mdlangchain-ai/langchainInstructions file149ktodayMITA4,680Codex, OpenCode
vscode oss-third-party-notices.instructions.mdmicrosoft/vscodeInstructions file195ktodayMITA5,516GitHub Copilot
vscode copilot-instructions.mdmicrosoft/vscodeInstructions file195ktodayMITA3,093GitHub Copilot
deepseek-harness AGENTS.mddeepseek-ai/deepseek-harnessInstructions file275k2d agoMITA4,313Codex, OpenCode