The fastest, cheapest, most efficient open-source agent harness. Run more than 500 agents on a $10 VPS.
Hooks
OpenHuman configures external hooks through the contracts owned by
tinyagents-hooks. The core keeps only the process-wide embedder registries in
agent/hooks.rs, implements HookIdentity for
OpenHumanRunContext, and installs the configured command HookEngine from
ops.rs.
Tool callbacks and their scoped registration live in tinyagents-hooks and
tinyagents-runtime. OpenHuman supplies its run context so a hook can inspect
the session, agent and working directory without introducing another hook
adapter. Post-turn learning uses the upstream LearningSink contract and
receives a privacy-preserving TurnSummary after the session commits.
The hooks.* RPC namespace and its model-facing command names remain stable.
Configuration parsing and command execution are supplied by
tinyagents_hooks::command; this module retains the OpenHuman controllers and
their lifecycle integration.