agentmods.dev

munin CLAUDE.md

A set of instructions for working on Munin, a customer platform whose features are accessed mainly by AI agents through MCP tools. It explains the repository structure and where shared business logic belongs.

In plain words
What is it for?
Use it when navigating Munin’s pnpm/Turborepo monorepo, locating backend modules, or changing its dashboard, API, MCP tools, widgets, and agent runtime.
Why use it?
It helps an agent understand the codebase before making changes, reducing the risk of putting logic in the wrong layer or breaking the shared API and MCP interfaces.

Instructions file

11 4d ago A scan 6,900 tokens original MIT
View the source file getmunin/munin
Install

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.

agentmods
npx agentmods add instructions/getmunin/munin/claude-md
Clone the repo
git clone --depth 1 https://github.com/getmunin/munin
Per session 6,900 This file is loaded in full into every session.
When invoked 6,900 The same file — it is already loaded in full.
Security scan A 0 findings. Scan, not verified.
Origin original No closer match found in the catalogue.
Token cost

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.

ModelPer sessionOnce invoked
Fable 5 $0.06900 $0.06900
Opus 5 $0.03450 $0.03450
Sonnet 5 $0.01380 $0.01380
Haiku 4.5 $0.00690 $0.00690

Measured 3d ago against content hash 18b90092d203, method: parsed. Prices are Anthropic first-party input rates as of 2026-08-30, from the pricing page.

Security

Grade A, and why

munin 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 3d 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.

CLAUDE.md · 203 lines

How it starts

The opening of the file, as written. The whole thing — 203 lines — stays where its author put it; the contents beside it link to each section on GitHub.

Munin — agent guide

MCP-first customer platform made for the agentic era (KB, Conversations, CRM, CMS, Outreach). The agent is the primary UI: agents act through MCP tools served at /mcp. The dashboard at apps/web is a thin shell over the /v1/* control plane; /v1 controllers and MCP tools must both stay thin wrappers over the same service methods — logic lives in the service, never in a controller or tool handler.

Repository layout

Monorepo on pnpm + Turborepo.

  • apps/backend — NestJS entry that composes @getmunin/backend-core with single-tenant AuthModule. Port 3001. Exposes /mcp, /auth/*, /.well-known/oauth-*, and the /v1/* control plane.
  • apps/web — Next.js 15 dashboard + landing. Port 3000. Calls apps/backend via /v1/*.
  • apps/chat-widget — embeddable browser bundle consumed via mn_widget_* keys.
  • packages/backend-core — shared NestJS modules. Where business logic lives.
  • packages/dashboard-pages — React pages reused by OSS and cloud webs.
  • packages/{core,db,types,sdk,mcp-toolkit,ui} — non-Nest building blocks.
  • packages/{agent-host,agent-runtime} — durable per-org LLM runner that drains the curator queue.
  • packages/widget-voice — Vapi voice glue.

Module map (packages/backend-core/src/modules/)

Module Tools prefix Notes
kb kb_* Documents, spaces, hybrid search (BM25 + pgvector), curation candidates.
conv conv_* Email / SMS / voice / widget channels via the channel-adapter contract (packages/backend-core/src/modules/conv/CLAUDE.md).
crm crm_* Contacts, companies, deals, pipelines, segments, merge proposals.
cms cms_* Collections, entries, locales, assets, scheduled publishing, public delivery API.
outreach outreach_* Propose-only outbound campaigns. Never auto-sends.
slack slack_* Operator bridge: mirrors conversations into Slack threads via a WebhookDispatcher sink + slack_deliveries queue. Not a ChannelAdapter.
connectors connectors_* Plumbing trunk for customer-facing data connectors: encrypted connection storage, vendor-adapter registry, credential testing. Domain modules register adapters here and own the typed read surfaces; live reads only, no vendor data persisted.
commerce commerce_* Connector-backed order lookups (Shopify, Magento 2): admin tools take an email; self-service tools bind to the calling end-user's email server-side.
bookings bookings_* Connector-backed bookings (Gastroplanner): read + write (availability, create/modify/cancel). Admin tools take an email; self-service tools bind to the calling end-user and enforce per-booking ownership.
seo seo_* Connector-backed search-console reads (Bing Webmaster Tools, Google Search Console) plus URL submission where the engine offers it. Admin-only — the one connector domain with no self-service half.
curator — Background job queue (curator_jobs) running skill://* and task://* URIs.
web — Website scraper (single task://web/scrape-website).
playbooks — Cross-module packaged workflows (skill markdown only, no tools).

Read the full file on GitHub · 203 lines

Changes

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.

  1. 3d ago First seen · 203 lines · 6,900 tokens per session scan A 18b90092d203

Subscribe to this mod's changes

munin CLAUDE.md is an instructions file published in the GitHub repository getmunin/munin (11 stars, last pushed 4d ago), licensed MIT. It adds 6,900 tokens to every session, about $0.0345 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.

Building agentschat-widgetcmscrmheadlesshubspot-alternative
Related

Other instructions, from other repositories

raisfast AGENTS.md

Instructions for RaisFast/raisfast, covering agents.md, project, commands, after changing the schema and postgres type discipline.

RaisFast/raisfast · 3,575 tokens

obscura AGENTS.md

Instructions for h4ckf0r0day/obscura, covering agents.md, build, rendering and stealth, no rendering, with rustls or stealth and test.

h4ckf0r0day/obscura · 2,630 tokens

payload AGENTS.md

Instructions for payloadcms/payload, a project described as: Payload is the open-source, fullstack Next.js framework, giving you instant backend superpowers. Get a full TypeScript backend and admin panel instantly. Use Payload as a headless CMS or for building powerful applications.

payloadcms/payload · 12 tokens

DeskcommCRM CLAUDE.md

Instructions for melgarafael/DeskcommCRM, covering claude.md — deskcommcrm, visão (1 parágrafo), stack canônica, convenções críticas (não negociáveis) and multi-tenancy.

melgarafael/DeskcommCRM · 11,401 tokens

DeskcommCRM AGENTS.md

Instructions for melgarafael/DeskcommCRM, covering agents.md — deskcommcrm, objetivo do projeto, stack (confirmado em package.json), estrutura que importa and comandos (confirmado em package.json).

melgarafael/DeskcommCRM · 5,114 tokens

microfeed AGENTS.md

AGENTS.md instructions for microfeed/microfeed, covering repository guidance, development workflow, instance management and cloudflare deployment, source architecture and theme repositories.

microfeed/microfeed · 2,529 tokens