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/crude-code/mcp-app/claude-mdgit clone --depth 1 https://github.com/crude-code/mcp-appWhat 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 | $0.10942 | $0.10942 |
| Opus 5 | $0.05471 | $0.05471 |
| Sonnet 5 | $0.02188 | $0.02188 |
| Haiku 4.5 | $0.01094 | $0.01094 |
Grade A, and why
mcp-app 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 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.
Makes network callslowCapability
Not a fault in itself. Listed so you know the mod talks to something, and to what.
spec)` / `fetch(user_slug, token)` — synchronous, spec always in hand at How it starts
The opening of the file, as written. The whole thing — 670 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Crude Code — MCP Server & Renderer
Oil & gas data analytics platform. MCP server + spec-driven inline renderer.
Architecture
Outer Claude does the thinking; the server does deterministic work. There are
no inner agents anymore — the managed-agents / inner-Opus / pip-package era was
removed in the rebuild. The one exception is get_skill: a static file bundle
(instructions + supporting files) Claude fetches and follows directly for
occasional, procedure-heavy tasks — not an agent, no code execution on the
server side. Outer Claude (in Claude Desktop / claude.ai) orchestrates
everything through a handful of MCP tools:
- It explores with
run_sql(direct, capped SELECT access) right in the chat. - When the chat becomes a deliverable, Claude itself builds a claude.ai
artifact straight from
run_sqldata — there is no server-side spec authoring or render step for this path anymore. - For deals it calls
deal_forecast_wells→deal_valuation→ gets a slim payload plus the frozenDealSheet.jsxtemplate in the same response, and builds the deal-sheet artifact from them directly. - For geography it calls
map_render. - For a one-off packaged procedure (e.g. extracting a dataroom upload) it
calls
get_skill(name)to fetch the instructions and follows them directly. - When the dataroom-extract skill produces an
extraction.json, it persists it viadataroom_save_extractionso the deal record outlives the chat. - When the user hits friction or wants something (a bug, a dataset request,
a feature wish) it files
message_team— durable row + best-effort email to the team.
Every tool is synchronous and server-side. The renderer today only ever
renders maps: it fetches the finished, hydrated map spec once via
map_read_full(token) and renders it inline — no streaming, no event log, no
polling.
MCP Server (server/mcp_server.py)
FastMCP server that Claude Desktop / claude.ai connect to. Plain Python tools —
no agent in the loop. Runs on port 9000 (/mcp endpoint). Per-user auth: a
reverse proxy routes https://<your-host>/<slug>/mcp to the server with an
X-User-Slug: <slug> header; identity resolves via the Supabase users table
(utils.platform.resolve_identity).
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 · 670 lines · 10,942 tokens per session scan A 3be898ff390a
mcp-app CLAUDE.md is an instructions file published in the GitHub repository crude-code/mcp-app (4 stars, last pushed 3d ago), licensed Apache-2.0. It adds 10,942 tokens to every session, about $0.0547 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 2026-08-31.
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.