Borrowing it
Nothing to install: this file belongs to susomejias/rembric. Take a copy, put it at the same path in your own repository, and replace the rules that are about this project with yours.
curl -O https://raw.githubusercontent.com/susomejias/rembric/main/CLAUDE.mdgit clone --depth 1 https://github.com/susomejias/rembricWrote 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/susomejias/rembric/claude-md)<a href="https://agentmods.dev/instructions/susomejias/rembric/claude-md"><img src="https://agentmods.dev/badge/instructions/susomejias/rembric/claude-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.
<a href="https://agentmods.dev/instructions/susomejias/rembric/claude-md"><img src="https://agentmods.dev/badge/instructions/susomejias/rembric/claude-md.svg" alt="Reviewed on agentmods" width="80" 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 5.1 | $0.04346 | $0.04346 |
| Opus 5 | $0.02173 | $0.02173 |
| Sonnet 5 | $0.00869 | $0.00869 |
| Haiku 4.5 | $0.00435 | $0.00435 |
Grade A, and why
rembric CLAUDE.md scanned grade A 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 6d 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.
Makes network callslowCapability
Not a fault in itself. Listed so you know the mod talks to something, and to what.
- **The TUI is the single install/maintenance path.** Install / setup / upgrade / uninstall of the server + every client plugin is done via the TUI installer (`apps/plugin/install.sh`, fronted by the repo-root `install.s How it starts
The opening of the file, as written. The whole thing — 106 lines — stays where its author put it; the contents beside it link to each section on GitHub.
CLAUDE.md
Guidance for Claude Code working in this repo. Specs in openspec/specs/ are the authoritative contract; this file is the must-know-fast index.
Quick reference
| Need | Command / path |
|---|---|
| Install | pnpm install (run corepack enable first) |
| Typecheck / lint | pnpm run typecheck · pnpm run lint |
| Tests | pnpm test (single file: pnpm vitest run path/to/file.test.ts) |
| Dev stack | pnpm run dev:docker:up (foreground, wipes+reseeds; docs/docker.md) |
| OpenSpec | /opsx:propose · /opsx:explore · /opsx:apply · /opsx:archive |
| Spec provenance | pnpm run check:spec-provenance (only origin/main...HEAD; CI is the gate) |
| Mutation check | node scripts/mutate.mjs --file … --spec … --mutation … --with … |
Conventional Commits required (commitlint). Pre-commit = lint-staged + tsc --noEmit --incremental. Pre-push = pnpm test. Never bypass git hooks with --no-verify.
PR titles and descriptions are always written in English (regardless of the language used in chat).
Operator surface = dashboard (/dashboard/{tokens,projects,sessions,judgments,memories,consolidation,maintenance}). No operator CLI; Docker image runs the server only.
Claims need evidence
A finding is not a finding until it has been executed. Scale the evidence to the cost of being wrong: the higher the consequence, the less an argument from reading the code is worth. This generalises the standard db-performance-auditor already applies to queries — measure the alternative, don't assume it.
- Never publish what you have not run. Issue comments, security advisories, PR descriptions and spec text are outward-facing; a wrong claim there has to be retracted in public.
- A subagent's finding is a hypothesis. Re-verify before acting on it or repeating it — subagents report plausible readings as results.
- Probe the boundary the real caller uses, and include a control that must pass. A probe that skips a layer proves nothing about it (calling an MCP handler directly bypasses its zod schema, so "the tool accepts X" measured that way is not a fact about the tool). With only a failing case you cannot tell a real defect from a broken probe; the control is what tells you which you have.
- Classify from behaviour, not from the symptom's shape. "The observable result changes" makes something a behaviour change; whether it is a defect depends on whether the current behaviour is defensible — same evidence bar as anything else.
- A new guard is not covered until its test fails without it. Weaken each condition separately and confirm the tests naming it go red —
scripts/mutate.mjsdoes the backup/mutate/run/restore loop. Three tests in one session passed while proving nothing (one asserted over an empty result set; two short-circuited before the condition they named), and only mutation found them. A test green on both sides of the change is the default outcome, not the exception. - One instrument per series, named. Never present an isolated statement's timing and an end-to-end operation's timing as one table: a 39× statement speedup was a 2.5× user-facing one, and the ratio came from silently switching instruments between rows. State which you measured, and quote the end-to-end figure when a user waits on it.
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.
- 6d ago Changed · -15 lines · -335 tokens per session 28c602522b29
- 10d ago First seen · 121 lines · 4,681 tokens per session scan A 36ad8266ad89
rembric CLAUDE.md is an instructions file published in the GitHub repository susomejias/rembric (12 stars, last pushed 7d ago), licensed MIT. It adds 4,346 tokens to every session, about $0.0217 per session on Opus 5. A static security scan graded it A with 1 finding (makes network calls). No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-30.
Other instructions, from other repositories
next.js AGENTS.md
AGENTS.md instructions for vercel/next.js, covering next.js development guide, codebase structure, monorepo overview, core package: packages/next and other important packages.
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 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.
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.