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/talonicdev/talonic-mcp/claude-mdgit clone --depth 1 https://github.com/talonicdev/talonic-mcpWrote 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/talonicdev/talonic-mcp/claude-md)<a href="https://agentmods.dev/instructions/talonicdev/talonic-mcp/claude-md"><img src="https://agentmods.dev/badge/instructions/talonicdev/talonic-mcp/claude-md.svg" alt="Measured on agentmods" 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 | $0.00660 | $0.00660 |
| Opus 5 | $0.00330 | $0.00330 |
| Sonnet 5 | $0.00132 | $0.00132 |
| Haiku 4.5 | $0.00066 | $0.00066 |
Grade A, and why
talonic-mcp 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 4d 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 — 34 lines — stays where its author put it; the contents beside it link to each section on GitHub.
CLAUDE.md — Claude Code entry point for talonic-mcp
Full agent orientation lives in AGENTS.md. Read it first — it covers the repo map, the eleven tools, the browser-handoff upload architecture, the CI/publish chain, and the common workflows. This file repeats only the two things you must not get wrong, plus the commands you'll use every session.
This repo is the Talonic MCP server (@talonic/mcp): document-extraction API exposed as MCP tools, running over stdio (local installs) and Streamable HTTP (mcp.talonic.com, for Claude.ai connectors).
⚠️ Before touching docs content — two parallel surfaces
src/content/sections/*.ts→@talonic/mcp/content→ renderstalonic.com/docs/mcp/*.docs/sections.json→@talonic/docs(synced via the platform repo) → becomes that package'smcpcontent domain, which nothing currently renders. It does not feed/docs/{sdk,api,platform}/*— those have their own content in the platform monorepo.
Edit the wrong one and your change silently never appears. docs/sections.json is maintained-but-dormant: the publish workflow's docs-drift guard still requires it to stay in step with src/tools/**, so keep it accurate, but user-visible MCP docs only come from src/content/sections/*.ts. The full map, the add-a-tool checklist, the failure-mode table, and the CI-token map are in docs/architecture/docs-pipeline.md. Read it before any non-trivial doc change.
⚠️ A push to main is a release
.github/workflows/publish.yml auto-bumps the version, publishes to npm + the MCP Registry, cuts a GitHub Release, and redeploys mcp.talonic.com. Don't push casually. The docs-drift guard fails the run if src/tools/**, src/http-server.ts, src/server-factory.ts, or src/resources/** changed without docs/sections.json — opt out with [skip docs] only for genuinely non-doc-affecting changes.
Quick commands
npm test # vitest, must pass before pushing
npm run typecheck # tsc --noEmit
npm run format:check # prettier check (npm run format to fix)
npm run build # tsup → dist/{index,server,http-server,content}.js
npm run start:http # local hosted-MCP server on :3000
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.
- 4d ago First seen · 34 lines · 660 tokens per session scan A 0415a2951eaa
talonic-mcp CLAUDE.md is an instructions file published in the GitHub repository talonicdev/talonic-mcp (5 stars, last pushed yesterday), licensed MIT. It adds 660 tokens to every session, about $0.0033 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
langgraph AGENTS.md
AGENTS.md instructions for langchain-ai/langgraph, covering agents instructions, corridor security analysis, libraries and dependency map.
ade-cli CLAUDE.md
Instructions for landing-ai/ade-cli, covering ade-cli, agent skills, issue tracker, triage labels and domain docs.
spacebot AGENTS.md
AGENTS.md instructions for spacedriveapp/spacebot, covering agents.md, what spacebot is, javascript tooling (critical), migration safety and delivery gates (mandatory).
cuga-agent AGENTS.md
AGENTS.md instructions for cuga-project/cuga-agent, covering agent guidelines, mark tests with their type, ci discovers tests by directory — do not edit workflow yaml, lazy imports — rules for new code and tests and cuga/init.py — pep 562 lazy exports.
template CLAUDE.md
Claude Code instructions for snowdreamtech/template, covering 🚨 critical system instruction 🚨, rules location, 🚨 critical token limit protocol 🚨 and why this file exists.
ai-harness CLAUDE.md
Claude Code instructions for samartomar/ai-harness, covering ai-harness — claude bootloader, vocabulary — two different things, read the source, not a summary of it, start here and self-hosting boundary.