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/agentregistry-dev/agentregistry/agents-mdgit clone --depth 1 https://github.com/agentregistry-dev/agentregistryWhat 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 | $0.03622 | $0.03622 |
| Opus 5 | $0.01811 | $0.01811 |
| Sonnet 5 | $0.00724 | $0.00724 |
| Haiku 4.5 | $0.00362 | $0.00362 |
Grade A, and why
agentregistry 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.
How it starts
The opening of the file, as written. The whole thing — 497 lines — stays where its author put it; the contents beside it link to each section on GitHub.
AGENTS.md - Development Guidelines
This document provides guidelines for both AI coding assistants (Claude, Copilot, Cursor, etc.) and human developers working on the AgentRegistry codebase.
Project Overview
AgentRegistry is a centralized registry for securely curating, discovering, deploying, and managing agentic infrastructure including MCP (Model Context Protocol) servers, agents, and skills.
Tech Stack:
- Backend/CLI: Go 1.26+
- Database: PostgreSQL (accessed via pgx)
- Frontend: Next.js 14 (App Router) with Tailwind CSS
- CLI Framework: Cobra
- API Framework: Huma (OpenAPI)
Architecture
Directory Structure
agentregistry/
├── cmd/ # Entry points only - minimal code
│ ├── cli/ # CLI entry point
│ └── server/ # Server entry point
├── pkg/ # Public, reusable packages
├── internal/ # Private implementation
│ ├── registry/ # Core registry implementation
│ │ ├── api/ # HTTP handlers
│ │ ├── database/ # Database layer (pgx)
│ │ ├── service/ # Business logic
│ │ └── ...
│ ├── cli/ # CLI command implementations
│ └── mcp/ # MCP protocol handling
├── ui/ # Next.js frontend
└── docker/ # Container images and local Compose configuration
Layer Responsibilities
- cmd/ - Entry points only. Delegate immediately to pkg/ or internal/
- pkg/ - Public APIs for external consumption and reusability
- internal/ - Implementation details, not importable by external packages
- internal/registry/database/ - ONLY place that accesses the database directly
- internal/registry/service/ - Business logic, receives database interface via constructor
Runtime Ownership
Deployment/runtime code should be organized by clear ownership, not by vague helper layers:
internal/registry/runtimes/<runtime>/owns runtime behavior - concrete packages such as kubernetes should contain their adapter plus the runtime-specific materialization/apply/discovery logic they needinternal/registry/runtimes/utils/is for narrowly shared deployment utilities only - use it for adapter-shared materialization helpers, validation, name generation, and request parsing that are truly cross-runtimeinternal/registry/runtimes/types/is for shared contracts only - keep shared schemas and DTO-style runtime types here, not behavior-heavy logicinternal/registry/api/handlers/is transport only - HTTP handlers should parse requests, call services/adapters, and map errors; they should not own deployment/runtime behaviorinternal/registry/registry_app.gois the composition root - wire concrete runtime adapters here explicitly instead of hiding registration/factory behavior in handler packages
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.
- 2d ago First seen · 497 lines · 3,622 tokens per session scan A 1fa5499bb210
agentregistry AGENTS.md is an instructions file published in the GitHub repository agentregistry-dev/agentregistry (478 stars, last pushed 4d ago), licensed Apache-2.0. It adds 3,622 tokens to every session, about $0.0181 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.
Other instructions, from other repositories
loom CLAUDE.md
Instructions for cosmix/loom, covering claude.md, project overview, build commands, architecture and source layout (loom/src/).
agentic-workflow CLAUDE.md
Instructions for gtrabanco/agentic-workflow, covering claude.md, repository layout, working rules, authoring a skill and hand off, don't compose across a model/effort boundary.
autovault CLAUDE.md
Instructions for autoworks-ai/autovault, covering claude.md, project, common commands, architecture and request flow.
agent-resources CLAUDE.md
Instructions for jcottam/agent-resources: See AGENTS.md for project conventions and contributing guidelines.
claude-for-strategy CLAUDE.md
Instructions for daddia/claude-for-strategy, covering claude.md, layout, validation — run before opening a pr, 2. json sanity and 3. consulting reference mirrors.
capgo-skills AGENTS.md
AGENTS.md instructions for Cap-go/capgo-skills, covering agents.md, quick start, install dependencies, build the plugin (typescript + rollup + docgen) and full verification (ios, android, web).