Overview
Agenthood is a full AI engineering team that earns every merge. Each of the 20 members has a defined role, trigger condition, and behavior — planners, builders, reviewers, auditors. They enforce engineering standards, review code and commits, manage memory, and run autonomously via the TypeScript runtime or inside Claude Code, with a tamper-evident audit trail for every decision.
Stack: TypeScript · Claude AI · Groq/OpenAI/Ollama/OpenCode/OpenCodeGo · LanceDB · Tree-sitter · VS Code
Links: agenthood.flabs.tech · GitHub · npm
Studio Playground
Chat with the whole team in your browser at agenthood.flabs.tech/studio/playground. SSE streaming, 7 providers, BYOK, server-side routing.
TypeScript Runtime
Run agents autonomously via the CLI with provider failover, circuit breaker, 5-tier memory, Agentic RAG, ReAct reasoning loops, and decision intelligence — every run records a provenance-backed, hash-chained decision in the audit trail.
Autonomy & Audit
Delegate a whole lifecycle — issue triage, planning, implementation, review, release — and every step is recorded: each member's decision carries a causal link and a tamper-evident hash chain, so autonomy never means "trust me." See ADR-015.
