llms.txt & AGENTS.md

llms.txt

Coldrune serves machine-readable documentation following the llmstxt.org convention:

• /llms.txt — curated index with links to key documentation pages and one-line descriptions
• /llms-full.txt — full documentation concatenated as a single plain text file

These files are designed for AI coding agents that need to understand Coldrune’s API, CLI, and configuration without parsing HTML.

When to use which

File	Use case
llms.txt	Quick orientation — what Coldrune does, where to find specific docs
llms-full.txt	Deep context — full documentation in one file for RAG indexing or large context windows

AGENTS.md for your projects

If your project uses Coldrune for secret management, add an AGENTS.md file to help AI agents work with your secrets correctly.

Template

# Secret Management

This project stores secrets in Coldrune.

## Configuration

- Config file: `coldrune.yml` in the project root
- Server: https://api.coldrune.com
- Organization: my-org
- Project: my-app

## Reading secrets

Secrets are available through the Coldrune MCP server or CLI:

- MCP: Use `secret_get` tool with the key name
- CLI: `coldrune secret get <KEY>`
- Runtime: `coldrune run -- <command>` injects all secrets as env vars

## Environments

- `dev` — local development
- `staging` — pre-production testing
- `prod` — production (read access requires explicit ACL grant)

## Rules

- Never log, print, or commit secret values
- Never hardcode secrets in source files
- Use `coldrune run` to inject secrets at runtime
- Use environment-specific secrets (don't share prod keys with dev)

Placement

Place AGENTS.md at the root of your repository. Most AI coding tools (Claude Code, Cursor, Copilot) read context files from the repository root automatically.

Tool	Context file
Claude Code	CLAUDE.md or AGENTS.md
Cursor	.cursor/rules/*.mdc or .cursorrules
GitHub Copilot	.github/copilot-instructions.md

You can add the Coldrune-specific section to your existing context file instead of creating a separate AGENTS.md.