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/dykstrasound/mcproxy/claude-mdgit clone --depth 1 https://github.com/DykstraSound/McProxyWhat 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 | $0.00975 | $0.00975 |
| Opus 5 | $0.00487 | $0.00487 |
| Sonnet 5 | $0.00195 | $0.00195 |
| Haiku 4.5 | $0.00097 | $0.00097 |
Grade A, and why
McProxy 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 — 67 lines — stays where its author put it; the contents beside it link to each section on GitHub.
McProxy — rules for working in this repo
Reasoning for every rule below lives in docs/ARCHITECTURE.md (section numbers noted). Read there
before changing behavior these rules protect.
Trust boundary (§1)
- No tool handler builds a
content: [...]literal. UseuntrustedResult/plainResult/errorResultfromsrc/shared/tools/untrusted.ts— nothing else. - Every mutating tool's description carries the literal all-caps word
MUTATES, and has a matching entry inAUDITED_MUTATING(scripts/test-mcp-trust.ts). The suite asserts the two sets are exactly equal, in both directions. - Provider error text reaches a user only through
sanitizeErrorMessage(src/shared/redact.ts, re-exported fromuntrusted.ts). Never surface a raw caught error.
.env (§3, §4)
src/connectors/env-file.tsis the only module that writes.env. Neverfs.writeFileit directly from anywhere else.envPrefixon a descriptor is always a literal you choose — never derive it fromid(an id may contain a hyphen, which isn't legal in an env var name).- The three naming forms (
{PREFIX}_ACCOUNTS, legacy adoption, bare singleton) are read in strict precedence and never merged. Don't try to make them cooperate.
Paths and roles (§7, §8)
- Never capture
baseDir()/tokenDir()in a module-levelconst/let/var. Both are functions — call them at the point of use, every time.scripts/test-boundaries.tscheck (c) scans for violations of this. - Every write under
~/.mcproxygoes throughsrc/shared/secure-fs.ts's helpers. Don't hand-rollfs.writeFile/fs.mkdiragainst that tree — the mode traps documented there are real.
Connectors (§2, §5, §6)
- A probe must never answer from a cache, and must never reject — encode every failure (missing
config, a rejected credential, a network error) into the returned
ProbeResult. - Identity is recorded only by an explicit act (Confirm identity, or a successful reconnect) — never inferred from an ordinary Test call.
- A descriptor may not claim more instance-plurality (
multiInstance: true) than its client actually implements. If the real client only ever holds one credential, the descriptor says so. - Absence is a state, not a default: no status record means "never checked," never
ok. Don't invent a default-healthy fallback anywhere in this layer.
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 · 67 lines · 975 tokens per session scan A c83931923dee
McProxy CLAUDE.md is an instructions file published in the GitHub repository DykstraSound/McProxy (0 stars, last pushed 20d ago), licensed MIT. It adds 975 tokens to every session, about $0.0049 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.
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.