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/komdosh/komnet/claude-mdgit clone --depth 1 https://github.com/Komdosh/komnetWrote 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/komdosh/komnet/claude-md)<a href="https://agentmods.dev/instructions/komdosh/komnet/claude-md"><img src="https://agentmods.dev/badge/instructions/komdosh/komnet/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 | $0.02458 | $0.02458 |
| Opus 5 | $0.01229 | $0.01229 |
| Sonnet 5 | $0.00492 | $0.00492 |
| Haiku 4.5 | $0.00246 | $0.00246 |
Grade B, and why
komnet CLAUDE.md scanned grade B 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 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.
Reads agent configuration directoriesmediumAgent snooping
.claude/, .codex/, .gemini/ hold keys, settings and other credentials a mod has no legitimate need for.
A user-level `~/.codex/config.toml` and `~/.gemini/settings.json` exist on this machine. To How it starts
The opening of the file, as written. The whole thing — 174 lines — stays where its author put it; the contents beside it link to each section on GitHub.
CLAUDE.md
This file provides guidance to Claude Code (claude.ai/code) when working with code in this repository.
Commands
pnpm verify # fmt:check + lint + build + test — this is exactly what CI runs
pnpm build # tsc --build (TypeScript 7 native compiler, project references)
pnpm test # builds first, then node --test over packages/*/test/**/*.test.ts
pnpm binary # → dist-bin/komnet, the self-contained executable
Running a subset:
node --test packages/core/test/core.test.ts # one file (build first)
node --test --test-name-pattern "converges when two agents" \
packages/core/test/core.test.ts # one test
pnpm build && node packages/cli/dist/bin.js --help # exercise the CLI directly
--test-name-pattern must come BEFORE the file path. After it, Node treats it as a
script argument and silently runs the whole file — reporting a pass, so nothing tells you
the filter was ignored. Always pass a file too: with no path it walks every suite, which
takes minutes.
pnpm test builds first on purpose: the CLI and MCP test suites spawn the built binary
(packages/cli/dist/bin.js), so a stale dist/ silently tests old code.
Requires Node 24+ (native TS execution, built-in node:sqlite) and git 2.42+
(worktree add --orphan).
What this is
A message bus for AI coding agents whose transport is a git repository. Rooms are git branches, messages are files, git history is the log. There is no server.
Read docs/design/00-north-star.md before making design decisions — it fixes the main idea
and everything else is downstream. spec/komnet-protocol-v1.md is the normative on-disk
contract; docs/adr/ records every significant decision with the alternatives rejected.
Invariants
These are not style preferences. Each holds up a load-bearing property, and breaking one
produces a bug that is very hard to trace back. CONTRIBUTING.md has the full rationale.
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 · 174 lines · 2,458 tokens per session scan B 533078022769
komnet CLAUDE.md is an instructions file published in the GitHub repository Komdosh/komnet (7 stars, last pushed 3d ago), licensed MIT. It adds 2,458 tokens to every session, about $0.0123 per session on Opus 5. A static security scan graded it B with 1 finding (reads agent configuration directories). No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-31.
Other instructions, from other repositories
relay AGENTS.md
AGENTS.md instructions for AgentWorkforce/relay, covering git workflow rules, never push directly to main, correct workflow, ... do work .. and stop here - let user merge.
PrismerCloud CLAUDE.md
Instructions for Prismer-AI/PrismerCloud, covering claude.md, what this is, sdk sync pipeline (critical mental model), sdk layout and build / test / release.
Bindu copilot.instructions.md
Copilot instructions for GetBindu/Bindu, covering bindu codebase patterns, tech stack, anti-redundancy rules, source of truth locations and settings (bindu/settings.py).
Bindu AGENTS.md
AGENTS.md instructions for GetBindu/Bindu, covering agents.md, repo orientation, useful commands, python core and run the core.
sema AGENTS.md
AGENTS.md instructions for emergent-wisdom/sema, covering repository agent guide, development workflow, vocabulary changes, vocabulary review and the improvement loop, per pattern.
Bindu CLAUDE.md
Claude Code instructions for GetBindu/Bindu, covering project: bindu - decentralized agent framework, critical context (read first), what is bindu?, commands that work and python core development.