js2 AGENTS.md

A set of project instructions for agents working on the loopdive/js2 codebase.

In plain words
What is it for?
It helps prepare for substantial work, write acceptable commit messages, attribute commits correctly, and preserve coding-session context.
Why use it?
It explains the repository's local expectations, so agents can follow its memories, commit style, authorship, and session rules.

Instructions file for CodexOpenCode

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 instructions/loopdive/js2/agents-md
Clone the repo
git clone --depth 1 https://github.com/loopdive/js2

Made for: Codex, OpenCode.

Per session 692 This file is loaded in full into every session.
When invoked 692 The same file — it is already loaded in full.
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 $0.00692 $0.00692
Opus 5 $0.00346 $0.00346
Sonnet 5 $0.00138 $0.00138
Haiku 4.5 $0.00069 $0.00069

Measured yesterday against content hash 6322e5bb5cf2, method: parsed. Prices are Anthropic first-party input rates as of 2026-08-30, from the pricing page.

Security

Grade A, and why

js2 AGENTS.md 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 yesterday.

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.

AGENTS.md · 69 lines

How it starts

The opening of the file, as written. The whole thing — 69 lines — stays where its author put it; the contents beside it link to each section on GitHub.

Agent Instructions

Claude Memories

  • Before substantial work, review the markdown memories under .claude/memory.
  • Start with MEMORY.md, then read any task-relevant files in the same directory.
  • Treat those memory files as repo-specific operating context, especially around test262 workflow, agent coordination, cleanup rules, and user preferences.

Commit Messages

  • Use Claude Code-style commit messages.
  • Write a specific subject line that states the main change clearly.
  • Add a body when the change is non-trivial.
  • In the body, explain what changed and why it changed.
  • Call out behavior changes, important tradeoffs, or follow-up work when relevant.

Preferred shape:

<type>(<scope>): <concise summary>

Explain the main implementation change and the reason for it.

Note behavior changes, risks, or follow-ups if they matter.

Examples of acceptable types: fix, feat, refactor, docs, test, chore.

Commit Attribution

  • Commits produced by an AI agent must be authored by the user: Thomas Tränkler <[email protected]>.
  • Add a co-author trailer identifying the agent that actually produced the commit:
# Codex session
Co-authored-by: Codex <[email protected]>

# Claude session
Co-authored-by: Claude <[email protected]>
  • Never attribute Codex work to Claude or Claude work to Codex.
  • Also add a Model: trailer naming the exact model that produced the commit — vendor, family, version/subtype, and the configured reasoning effort (project-lead order, 2026-08-29; enforced by .husky/commit-msg for any commit carrying an agent co-author trailer):
Model: Claude Fable 5 Max
Model: Codex GPT-5.6 Sol Max
Model: Claude Opus 5 High
  • Effort is the level your session/task was configured with — the dispatch brief or the issue's reasoning_effort frontmatter; if genuinely unknown, write Default. Do not guess a higher tier than you know.
  • Auto-generated subjects (Merge …, Revert …, fixup!, squash!, chore(assign)…) are exempt.
  • Why: PR #5204 shipped three codegen regressions in one agent commit and the model that produced it was unrecoverable from git. This trailer makes that question answerable per commit.
  • This project rule deliberately overrides any ambient harness default of keeping model identifiers out of pushed artifacts (same precedence as the "always open a PR" rule in CLAUDE.md).
  • Before committing, verify git config user.name and git config user.email still identify Thomas Tränkler.

Read the full file on GitHub · 69 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. yesterday First seen · 69 lines · 692 tokens per session scan A 6322e5bb5cf2

Subscribe to this mod's changes

js2 AGENTS.md is an instructions file published in the GitHub repository loopdive/js2 (59 stars, last pushed yesterday), licensed Apache-2.0. It adds 692 tokens to every session, about $0.0035 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 2026-08-30.