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.
npx agentmods add instructions/rordi-ai/loopbreaker/agents-mdgit clone --depth 1 https://github.com/rordi-ai/loopbreakerWrote 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.
[](https://agentmods.dev/instructions/rordi-ai/loopbreaker/agents-md)<a href="https://agentmods.dev/instructions/rordi-ai/loopbreaker/agents-md"><img src="https://agentmods.dev/badge/instructions/rordi-ai/loopbreaker/agents-md.svg" alt="Measured on agentmods" height="20"></a>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 | $0.00610 | $0.00610 |
| Opus 5 | $0.00305 | $0.00305 |
| Sonnet 5 | $0.00122 | $0.00122 |
| Haiku 4.5 | $0.00061 | $0.00061 |
Grade A, and why
loopbreaker 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 4d 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.
What it actually says
Loopbreaker agent guide
Loopbreaker is an agent-facing CLI and local MCP server. Keep machine output compact and stable. CLI stdout is TOON; diagnostics belong on stderr.
Before completing a change, run:
pnpm verify
This also verifies the seven plugin skills, both plugin manifests (Codex and
Claude Code), the bundled MCP entry point, tool discovery, and one real MCP call.
When the Codex skill/plugin validators are installed, also run their validators
against each skills/* directory and the repository root.
For the live demo, initialize a disposable database and exercise both surfaces:
loopbreaker demo --db /tmp/loopbreaker-demo.db
loopbreaker substrate DEMO-1 --db /tmp/loopbreaker-demo.db
loopbreaker serve --db /tmp/loopbreaker-demo.db
The domain rules are non-negotiable:
- A shape cannot reach
proceedwithout a founder-approved discovery record. Discovery is the first ordered authority; the premise must come from a human. - Behavior children are enforced by default unless explicitly advisory.
- An issue's behavior children freeze its acceptance contract.
- Implementation requires shape
proceed, healthy planning, and independent planning-review approval in that order. - Planning review and implementation review each use at most three passes: comprehensive, repair verification, and decision only. Neither has pass four.
- Three passes is a budget, not a verdict on the work. An implementation review
that spends all three without passing ends that ROUND, not the issue: the open
findings are the work list, and
loopbreaker review-roundopens a new round that starts again at pass one. The loop stays broken because opening a round is a human act — it needs a name and a reason, it is refused while the current round still has a pass left, and every round is on the record. What is refused is a fourth pass grafted onto a review that already concluded, or a silent ship past findings nobody repaired. - Review completion and shipping readiness are separate facts.
- Shipping requires every enforced behavior to be verified or explicitly waived.
- Evidence is executed, never asserted. An enforced behavior can only be verified
on a proof loopbreaker ran itself (
loopbreaker prove), whose verdict comes from the harness exit code.not_runis the fail-closed default: failing to observe a pass is never the same as passing. - A behavior's harness is bound in two directions. The behavior names a registry
entry (
loopbreaker bind, outside the contract freeze) and the entry names the behavior back in itsproveslist (a reviewed change toharnesses.json). Neither half alone is sufficient. - Build a behavior's harness and prove it RED against current HEAD before implementing. A harness that is green before the work exists proves nothing.
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.
- 4d ago First seen · 55 lines · 610 tokens per session scan A 1e2766a46201
loopbreaker AGENTS.md is an instructions file published in the GitHub repository rordi-ai/loopbreaker (3 stars, last pushed 1mo ago), licensed MIT. It adds 610 tokens to every session, about $0.0030 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.
Other instructions, from other repositories
vscode buildNext.instructions.md
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).
spec-kit AGENTS.md
AGENTS.md instructions for github/spec-kit, covering agents.md, about spec kit and specify, quickstart — add a new integration in 5 steps, integration architecture and integrationmanifest — file tracking.
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.
vscode oss-third-party-notices.instructions.md
Instructions for microsoft/vscode, covering vs code oss third-party-notices pipeline, architecture, pipeline flow in ci, applying the notice (cutover) and fallback chain (never fail the build).
langchain AGENTS.md
AGENTS.md instructions for langchain-ai/langchain, covering global development guidelines for the langchain monorepo, corridor security analysis, project architecture and context, monorepo structure and development tools & commands.
deepseek-harness AGENTS.md
AGENTS.md instructions for deepseek-ai/deepseek-harness, covering agents.md, pre-release stance: foundation over blast radius, repository layout, commands and host sandbox failures.