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/falldownthesystem/converse/claude-mdgit clone --depth 1 https://github.com/FallDownTheSystem/converseWrote 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/falldownthesystem/converse/claude-md)<a href="https://agentmods.dev/instructions/falldownthesystem/converse/claude-md"><img src="https://agentmods.dev/badge/instructions/falldownthesystem/converse/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 5.1 | $0.03867 | $0.03867 |
| Opus 5 | $0.01934 | $0.01934 |
| Sonnet 5 | $0.00773 | $0.00773 |
| Haiku 4.5 | $0.00387 | $0.00387 |
Grade A, and why
converse 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 5d 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 — 421 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Claude Development Guide for Converse MCP Server
This repository contains the Converse MCP Server - a functional Node.js implementation of an MCP server.
Repository Structure
src/- Main source code for the Converse MCP Serverdocs/- Complete documentation (API, Architecture, Examples)tests/- Comprehensive test suite (unit, integration, e2e)scripts/- Development and build scriptsexamples/- Usage examples and sample configurationsbin/- Executable binaries for CLI usage
Converse MCP Server (Node.js)
The Converse MCP Server follows a functional architecture with a single unified chat tool that takes a mode parameter:
chatmode - 1..N models answer independently in parallelconsensusmode - ≥2 models answer in parallel, then refine via cross-feedbackroundtablemode - models answer sequentially, each building on the running transcript
Development Setup
IMPORTANT: This project uses pnpm, not npm.
# Install dependencies (use pnpm, NOT npm)
pnpm install
# Start development server
pnpm run dev
# Run tests
pnpm test
For detailed implementation guidance, see:
docs/API.md- Complete API referencedocs/ARCHITECTURE.md- System architecture and design principlesdocs/EXAMPLES.md- Usage examples and patterns
Release & CI Automation
This repo is fully automated for releases via release-please. Do not manually bump versions in package.json, manually run npm publish, or manually edit the top of CHANGELOG.md — the automation owns these.
How releases work
- Every push to
mainwith a user-facing commit triggers a release. release-please skips pushes whose commits are all hidden types (ci:,chore:,docs:, etc.) with "No user facing commits found — skipping"; force one of these with aRelease-As:footer if you need to. Dependabot auto-merges explicitly dispatch the same workflow after merge so dependency bumps still publish a patch release (always-bump-patchstrategy). Workflow:.github/workflows/release.yml. - The flow in a single workflow run (~50s):
- release-please opens (or updates) a "chore(main): release X.Y.Z" PR with version bump + CHANGELOG.md entry.
- Workflow auto-merges that PR with
gh pr merge --squash. - release-please runs again, sees the just-merged release PR, tags
vX.Y.Z, creates a GitHub Release. - Workflow runs
pnpm install --frozen-lockfile+ lint + typecheck + test:unit, upgrades npm (npm install -g npm@latest), thennpm publish --access public.
- Branch protection on
mainis off — direct pushes are allowed and intentional, since the automation needs to push the auto-merge commit. - Publishing uses npm Trusted Publishing (OIDC) — no
NPM_TOKENsecret. The job'sid-token: writepermission + a trusted publisher configured on npmjs.com (orgFallDownTheSystem, repoconverse, workflowrelease.yml) authenticate the publish, and provenance is generated automatically. Trusted Publishing requires npm >= 11.5.1, which is why the workflow upgrades npm before publishing.
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.
- 5d ago First seen · 421 lines · 3,867 tokens per session scan A 7ca5107359a8
converse CLAUDE.md is an instructions file published in the GitHub repository FallDownTheSystem/converse (2 stars, last pushed yesterday), licensed MIT. It adds 3,867 tokens to every session, about $0.0193 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
openrouter-deep-research-mcp CLAUDE.md
Claude Code instructions for wheattoast11/openrouter-deep-research-mcp, covering agent zero: isomorphic protocol bridge (v2.0.0), system identity, operational paradigms, database persistence (v2.0.0+) and core commands.
openrouter-deep-research-mcp AGENTS.md
AGENTS.md instructions for wheattoast11/openrouter-deep-research-mcp, covering agents.md, isomorphic 5-layer architecture alignment, layer correspondence (axon architecture), resilience & production status and quick commands.
mcp-tool-kit CLAUDE.md
Instructions for getfounded/mcp-tool-kit: Your job is to first clean up this repository and make a more dynamic ragistration system for the existing tools, clean up the tool files, clean up the repo in general, implement local sse, so the server can be ran via sse transport. Try to make an easier way for people to…
ida-pro-mcp CLAUDE.md
Instructions for mrexodia/ida-pro-mcp, covering claude.md, what this project is, core implementation rules, ida thread safety and api conventions.
tableau-mcp CLAUDE.md
Claude Code instructions for tableau/tableau-mcp, covering claude.md, build & development commands, building, running the server and testing.
GhidrAssistMCP CLAUDE.md
Claude Code instructions for symgraph/GhidrAssistMCP, covering claude.md, build and development commands, building the extension, set ghidrainstalldir environment variable first and build the extension.