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/ncr/rack-proxy/claude-mdgit clone --depth 1 https://github.com/ncr/rack-proxyWhat 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 | $0.01584 | $0.01584 |
| Opus 5 | $0.00792 | $0.00792 |
| Sonnet 5 | $0.00317 | $0.00317 |
| Haiku 4.5 | $0.00158 | $0.00158 |
Grade A, and why
rack-proxy CLAUDE.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.
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.
Working in this repo (for AI agents and new contributors)
rack-proxy is a small, security-sensitive Rack middleware/app that proxies HTTP
requests to a backend and lets you rewrite the request and response. It is a
library that other apps mount — so the bar is: safe defaults, no surprises, and
never regress the security behavior below. Read this before changing code.
Commands
bundle install
bundle exec rake test # full suite, fully OFFLINE, ~2-3s
LIVE=1 bundle exec rake test # additionally runs real-internet smoke tests
bundle exec standardrb # style check, CI-blocking (--fix to autofix)
COVERAGE=1 bundle exec rake test # SimpleCov, ratcheted floor (CI-enforced)
# Run the suite against a specific Rack major (CI does both):
BUNDLE_GEMFILE=gemfiles/rack_3.gemfile bundle exec rake test
BUNDLE_GEMFILE=gemfiles/rack_2.gemfile bundle exec rake test
The default suite must never touch the network. CI (.github/workflows/ci.yml)
runs the matrix Ruby 3.1–3.4 × Rack 2/3, a ruby head canary, a gem-build
smoke test, and a blocking lint job (standardrb + coverage floor + bundler-audit).
Architecture (two core files, ~460 lines)
lib/rack/proxy.rb—Rack::Proxy. The entry point:call→rewrite_env→perform_request→rewrite_response.perform_requestextracts/forwards request headers, picks the backend, and has two distinct network paths that must stay behaviorally identical: streaming (default) and non-streaming (streaming: false). Subclasses overriderewrite_env/rewrite_response(and sometimesperform_request); seeexamples/(copy-paste snippets, not shipped in the gem).lib/rack/http_streaming_response.rb—HttpStreamingResponse, the lazy Rack body used by the streaming path. It runs the public block form ofNet::HTTP#requestinside a Fiber: the Fiber pauses once the status and headers are in, and#eachresumes it to pull body chunks;#each/#closetear the connection down (early termination unwinds the Fiber viaFiber#raise).lib/net_http_hacked.rb— the former streaming engine (a monkey-patch of privateNet::HTTPinternals) — was deleted in 1.0. Do not reintroduce it or anything like it; a subprocess test asserts the library loads without defining the old hacked methods.
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 · 106 lines · 1,584 tokens per session scan A 1156552a2a4c
rack-proxy CLAUDE.md is an instructions file published in the GitHub repository ncr/rack-proxy (284 stars, last pushed 1mo ago), licensed MIT. It adds 1,584 tokens to every session, about $0.0079 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.
Other instructions, from other repositories
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 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).
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.
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.
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).
next.js AGENTS.md
Instructions for vercel/next.js, covering next.js development guide, codebase structure, monorepo overview, core package: packages/next and other important packages.