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/clay-good/roughlogic.com/agents-mdgit clone --depth 1 https://github.com/clay-good/roughlogic.comWhat 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 | $0.01028 | $0.01028 |
| Opus 5 | $0.00514 | $0.00514 |
| Sonnet 5 | $0.00206 | $0.00206 |
| Haiku 4.5 | $0.00103 | $0.00103 |
Grade A, and why
roughlogic.com 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 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 — 75 lines — stays where its author put it; the contents beside it link to each section on GitHub.
AGENTS.md
Guidance for AI agents working in or with this repository. (Humans: see README.md; the MCP server has its own mcp/README.md.)
What this repo is
roughlogic.com is a static, offline-first site of 1,804 calculators for the trades (electrical, plumbing, HVAC, construction, restoration, and more) across 57 calc modules, plus a local, zero-cost MCP server that exposes every one of them to an AI agent. US standards only. No AI at runtime. Calculator execution is local; the only hosted write path is the bounded, user-initiated D1 feedback endpoint, and the MCP server runs on your machine over stdio.
Use the calculators as an agent (the fast path)
The server is local, zero-dependency, Node 18+. Point any MCP client at it:
claude mcp add roughlogic -- node /absolute/path/to/roughlogic.com/mcp/server.mjs
Four tools, all read-only and safe to call freely:
search_calculators— find a calculator by keyword and/or trade (no args → trade overview).describe_calculator— one tile's inputs (with select options, units, min/max), outputs, worked examples, citation, and any limitation banner.run_calculator— evaluate a tile; returns the raw result plus rendered outputs (units + display strings), range warnings, and the limitation banner.answer_query— a plain-language question with its numbers in it, answered in one call. Use it instead of chaining search + describe + run when the question already carries its values.run_calculators— up to 50{ id, inputs }calls in one request, for sweeps and comparisons.
Typical flow: search_calculators → describe_calculator → run_calculator. The
server also serves MCP resources (roughlogic://catalog, roughlogic://trade/{trade},
roughlogic://calculator/{id}) and prompts (find-calculator, run-with-inputs,
size-and-check). See mcp/README.md for client-specific config.
The repo map
| What | Where |
|---|---|
| Compute functions and renderers | calc-*.js (one module per trade group) |
| Tile registry (id → name, group, trades) | tools-data.js |
| Tile id → compute function | test/fixtures/compute-map.js |
| Tile id → renderer (for field schemas) | test/fixtures/renderer-map.js (generated from tool-modules.js) |
| Publisher-verified worked examples | test/fixtures/worked-examples.json |
| MCP server + catalog layer | mcp/server.mjs, mcp/catalog.mjs |
| Shared calculator report UI | report-feedback.js |
| Defensive report Worker + D1 | report-worker.mjs, migrations/ |
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 · 75 lines · 1,028 tokens per session scan A 01ff1331b8b9
roughlogic.com AGENTS.md is an instructions file published in the GitHub repository clay-good/roughlogic.com (3 stars, last pushed 3d ago), licensed MIT. It adds 1,028 tokens to every session, about $0.0051 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
opentakeoff AGENTS.md
Instructions for Kentucky-ai/opentakeoff, covering run / build / check, shipping — the required steps, every change, where things live, how the canvas works (the mental model) and conventions.
massing AGENTS.md
Instructions for ibuilder/massing, covering agents.md, cursor cloud specific instructions, what the update script already did, services (all run from source in dev) and node 24 gotcha (important).
massing CLAUDE.md
Instructions for ibuilder/massing, covering project: massing, what this is, non-negotiables, stack and build order.
step.parts AGENTS.md
Instructions for earthtojake/step.parts, covering agent guide, quick commands, repo shape, adding parts and metadata rules.
procore-mcp-server CLAUDE.md
Instructions for TylerIlunga/procore-mcp-server, covering procore mcp server, quick start, architecture, build pipeline and tool description quality.
cpp-cpm-engine CLAUDE.md
Claude Code instructions for danafitkowski/cpp-cpm-engine, covering claude.md — operating contract for cpp-cpm-engine releases, 1. definition of done (dod) — the four criteria, 2. status taxonomy (mirrored in every ledger), 3. hard-stop conditions before any release and 4. operating rules for ledgers.