Everything load-bearing, in one place.
Runelm ships with a small, opinionated documentation surface. Four pages cover the architecture, the API, the threat model, and the comparison landscape. The same pages are linked from the top nav — this index just gives you the map.
Reference
- ᚨ
Architecture
The nine-stage classification pipeline, the routing matrix, the four design principles. Start here if you want the technical reference.
Read → - ᛟ
API
Three install patterns — proxy, decorator, direct API. Proxy headers, runelm.yaml config, CLI surface.
Read → - ᛇ
Threat model
What Runelm defends against, what it does not, and the auditor-facing stakeholder template for the MEDIUM path's contractual residual risk.
Read → - ᛞ
Comparison
Where each load-bearing property sits across the OSS landscape — what we learned from, what we deliberately did not copy.
Read → - ᚷ
Integrations
Maintained adapters for LiteLLM, LangChain, and the OpenAI Agents SDK — wrap the model in place, keep the fail-closed guarantee.
Read → - ᛉ
Operations
The production surface: observability dashboard, dry-run replay, rate limiting, tamper-evident audit, metrics, and a hardened container.
Read →
Want something that is not here?
File an issue on GitHub or send us a note. Documentation gaps get written up the same week they are reported.