moss AGENTS.md

Repository instructions for Moss, a local semantic-search system that helps AI agents find related text quickly. They describe the project, its folders, SDKs, integrations, and examples.

In plain words
What is it for?
Use them when working on Moss SDKs, framework integrations, application examples, or cookbook code in Python, JavaScript, Elixir, Go, C, and related environments.
Why use it?
They give coding agents the project context and conventions needed to make changes in the right part of the repository.

Instructions file for CodexOpenCode

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/usemoss/moss/agents-md
Clone the repo
git clone --depth 1 https://github.com/usemoss/moss

Made for: Codex, OpenCode.

Per session 4,222 This file is loaded in full into every session.
When invoked 4,222 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 $0.04222 $0.04222
Opus 5 $0.02111 $0.02111
Sonnet 5 $0.00844 $0.00844
Haiku 4.5 $0.00422 $0.00422

Measured 2d ago against content hash bbab9974ab47, method: parsed. Prices are Anthropic first-party input rates as of 2026-08-30, from the pricing page.

Security

Grade A, and why

moss 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.

AGENTS.md · 287 lines

How it starts

The opening of the file, as written. The whole thing — 287 lines — stays where its author put it; the contents beside it link to each section on GitHub.

AGENTS.md

This file provides guidance to AI agents when working with code in this repository.

What This Repo Is

Moss is a real-time semantic search runtime for AI agents targeting sub-10ms query latency. The runtime runs on-device using bundled embedding models (moss-minilm) no external embedding API calls are needed. The cloud layer handles project management and index distribution; the local layer handles querying.

This repository contains the multi-language SDKs, framework integrations (cookbooks), and application examples.

Repository Layout

sdks/
  python/sdk/      — Python SDK (PyPI: moss), Python 3.10+
  javascript/sdk/  — JS/TS SDK (npm: @moss-dev/moss), ESM-only
  elixir/sdk/      — Elixir SDK (Hex: moss)
examples/
  python/          — Standalone Python usage examples
  javascript/      — Standalone TS usage examples
  javascript-web/  — Browser/Vite examples (no Node runtime)
  c/               — C binding examples
  go/              — Standalone Go SDK usage examples
  bun/             — Bun runtime example
  python-classification/ — Text classification with Moss
  voice-agents/    - End-to-end voice agents (LiveKit-based)
    airline-pnr/         - Ambient retrieval; per-caller Moss index, swap mid-call
    mortgage-lending/    - Multi-agent flow with shared session state
  cookbook/        — Framework integrations (one subdirectory per framework)
    autogen/       — AutoGen multi-agent e-commerce support
    crewai/        — CrewAI retrieval tool
    daytona/       — Log Ingestion Q&A Agent on Daytona sandboxes
    dspy/          — DSPy notebook
    haystack/      — Haystack RAG pipeline integration
    langchain/     — LangChain retriever + tool integration
    langflow/      — Langflow drag-and-drop retriever + search components
    langgraph/     - LangGraph stateful retrieval node
    mastra/        — Mastra agent createTool() integration
    pydantic-ai/   - Pydantic AI integration
    moss-cognee-daytona/ — Claude Code + Cognee + Moss on Daytona (shared memory)
apps/
  agora-moss/      — Agora Conversational AI voice agent (MCP server demo)
  agora-custom-llm-moss/ — Agora custom-llm middleware: ambient prepend + in-process tool loop
  docker/          — Dockerized Python + JS SDK examples (ECS/K8s pattern)
  elevenlabs-moss/ — ElevenLabs voice agent with Moss knowledge base
  livekit-moss-vercel/ — LiveKit voice agent + React frontend on Vercel
  moss-bun/        — Production Bun semantic search application
  moss-llamaindex/ — LlamaIndex + Liteparse full-stack PDF search demo
  next-js/         — Next.js 16 browser-based semantic search UI (@moss-dev/moss-web)
  pipecat-moss/    — Pipecat voice agent (three variants below)
    pipecat-quickstart/  — Cloud-deployable quickstart bot
    ollama-local/        — Local LLM + Moss + Pipecat via docker compose
    hume-ollama-local/   — Local LLM + Hume AI TTS + Moss + Pipecat
  ten-moss/        — TEN Framework voice agent with Moss session-scoped grounding
  vapi-moss/       — VAPI Custom Tool webhook server
packages/
  agora-moss/            — Agora Conversational AI MCP server package
  elevenlabs-moss/       — ElevenLabs integration package
  moss-cli/              — CLI for index/document management (no-code workflows)
  moss-data-connector/   — Database source connectors
    moss-connector-mongodb/  — MongoDB connector
    moss-connector-mysql/    — MySQL / MariaDB connector
    moss-connector-sqlite/   — SQLite connector
    moss-connector-supabase/ — Supabase (PostgREST) connector
  moss-md-indexer/       — Markdown docs → Moss index builder
  pipecat-moss/          — Pipecat Python integration package
  strands-agents-moss/   — AWS Strands Agents integration package
  ten-moss/              — TEN Framework Moss session manager (MossSessionManager)
  vapi-moss/             — VAPI Custom Knowledge Base webhook adapter
  n8n-nodes-moss/        — n8n community node for Moss index + query
  vercel-sdk/            — Vercel AI SDK tool wrappers (@moss-tools/vercel-sdk)
  vitepress-plugin-moss/ — VitePress search plugin (on-device fallback after cloud)
  zo-computer/           — Zo computer skill for Moss search
moss-live-labs/          - Experimental zone: prototypes and community demos (APIs can change)
  python/                - Minimal Python quickstart + advanced query example
  typescript/            - Minimal TypeScript quickstart + advanced query example
  examples/
    voice-agent/         - LiveKit + Moss voice assistant
    advanced-voice-agent/ - Persona impersonator built on a PDF knowledge base
    image-search/        - FastAPI + React image search over COCO data
  community-demos/
    voice-agents/
      bharat-benefits/      - Voice RAG over Indian public-benefit schemes (Sarvam STT/TTS)
      shoplabs-voice-agent/ - Pipecat WebRTC ecommerce support agent

Read the full file on GitHub · 287 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. 2d ago First seen · 287 lines · 4,222 tokens per session scan A bbab9974ab47

Subscribe to this mod's changes

moss AGENTS.md is an instructions file published in the GitHub repository usemoss/moss (669 stars, last pushed 9d ago), licensed BSD-2-Clause. It adds 4,222 tokens to every session, about $0.0211 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 2026-08-30.