The top-level README is the high-level intro; everything below is for people (and agents) who need to go deeper.
The full technical manual for hamroh itself. Read this when you're modifying, debugging, or auditing the project.
Covers: every env var, the four-process architecture in detail, how to add tools and skills, access control internals, memory + reminders, the four monitoring windows (live log, session replay, raw wire log, SQLite), the complete security model with all invariants, the manual end-to-end checklist, and the full repo layout.
The canonical list of every tool available to the bot — always-on
hamroh built-ins, always-on Claude Code built-ins (web), and
opt-in groups (subagents, shell, code, Jira, GitLab, GitHub). Also
the canonical reference for plugins.json: the
schema, how to plug in a new external MCP, how to disable a built-in
tool you don't use, and how to hide a skill. Read this when deciding
what capabilities to grant for your deployment.
Step-by-step guide for deploying hamroh to a VPS (Hetzner, DigitalOcean, Contabo…) using Docker, plus a continuous-deployment workflow. Read this when you're moving the bot from your laptop to a server, or wiring it into CI.
Notes on the two systems hamroh descends from: Anthropic's official Telegram plugin and the Rust Claudir. Read this before proposing architectural changes — it explains which patterns came from where and why, so you don't accidentally regress to a shape that's already been ruled out.
| You want to… | Read |
|---|---|
| Run the bot locally | ../README.md |
| Decide which tools to enable | tools.md |
| Understand a specific env var or security rule | documentation.md |
| Deploy to a server | deployment.md |
| Propose a structural / architectural change | reference-architectures.md then documentation.md |