What is loopkit AGENTS.md?

A session guide for Loopkit, a repository that coordinates long-running coding-agent sessions through a repeated plan, act and verify cycle. It requires each session to deliver one feature and end in a clean, checked state.

Use it to plan one repository change, implement it, run the project's verification review, check disagreements with the plan, and leave the work ready for the next session.

It prevents agents from taking on too much work at once or declaring a change finished without testing it. It also explains which project files and recent Git history to inspect first.

loopkit AGENTS.md — Instructions file by Archive228

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/archive228/loopkit/agents-md

Clone the repo

git clone --depth 1 https://github.com/Archive228/loopkit

Made for: Codex, OpenCode.

Per session 1,078 This file is loaded in full into every session.

When invoked 1,078 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.

Model	Per session	Once invoked
Fable 5	$0.01078	$0.01078
Opus 5	$0.00539	$0.00539
Sonnet 5	$0.00216	$0.00216
Haiku 4.5	$0.00108	$0.00108

Measured 2d ago against content hash b6ea8b9b9cc7, method: parsed. Prices are Anthropic first-party input rates as of 2026-08-30, from the pricing page.

Security

Grade A, and why

loopkit 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 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.

AGENTS.md · 72 lines

How it starts

The opening of the file, as written. The whole thing — 72 lines — stays where its author put it; the contents beside it link to each section on GitHub.

AGENTS.md — loopkit loop contract

This repo runs long-lived agent sessions. Every session shares one contract: Plan → Act → Verify, single-feature per session, clean state at the end.

This file is the cross-agent voice (Claude Code, Cursor, Codex CLI, Gemini CLI, Amp). Claude-specific extras live in .claude/CLAUDE.md, which imports this file via @../AGENTS.md.

The three-step loop

Every session, in order:

Plan. Read PROMPT.md (goal), IMPLEMENTATION_PLAN.md (state), and git log --oneline -20 (history). If the last session claimed a feature done, smoke-test it before picking new work.
Act. Implement exactly one feature. Not two. Not "one and a small one".
Verify. Run /verify (adversarial pass against the diff) BEFORE claiming done or committing. Non-zero from /verify blocks the commit.

If IMPLEMENTATION_PLAN.md and the git log disagree, trust the git log. Git is append-only; the plan is rewritten each turn.

Single-feature rule

One feature per session. Sessions that pack multiple features ship them all half-done — we have measured this. The next session takes the next feature. There is no reward for finishing more in one session; there is a real cost to leaving things half-done.

Clean-state contract (end of every session)

All code committed to git. Prefer scripts/committer "<msg>" <files> — it refuses . and empty messages.
No uncommitted changes in the working tree.
IMPLEMENTATION_PLAN.md updated: what was done, what is next, known open issues.
Dev server killed (./stop.sh if the project has one).
A feature is only "done" after end-to-end verification (/verify), not after unit tests alone.

Skills vs rules

Skills (skills/*/SKILL.md) — invoked on demand by trigger phrases in their description. Read the skill's SKILL.md before acting on a matching task.
Rules (.claude/rules/*.md) — auto-loaded when a file path matches. Silent guardrails, not opt-in.

Full skill routing table: skills/using-loopkit/SKILL.md (repo layout) or .claude/skills/using-loopkit/SKILL.md (installed layout).

Read the full file on GitHub · 72 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.

2d ago First seen · 72 lines · 1,078 tokens per session scan A b6ea8b9b9cc7

Subscribe to this mod's changes

loopkit AGENTS.md is an instructions file published in the GitHub repository Archive228/loopkit (753 stars, last pushed 1mo ago), licensed MIT. It adds 1,078 tokens to every session, about $0.0054 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.

codex AGENTS.md

AGENTS.md instructions for openai/codex, covering rust/codex-rs, the codex-core crate, code review rules, crate api surface and model visible context.

openai/codex · 5,182 tokens

buildNext

Working notes and architecture documentation for the new esbuild-based build system in build/next. Use when making changes to the new build pipeline (transpile/bundle commands, NLS plugin, source-map handling, resource copying, or self-hosting watch tasks).

microsoft/vscode · 6,785 tokens

loopkit AGENTS.md