brain

Give an agent a persistent, evidence-first "brain" — a git repo it records experiences into and recalls grounded, cite-or-abstain answers from, with a deterministic constraint shield that vetoes "profitable but ruinous" decisions. Backed by the standalone `brain` CLI (Go, no network, no model endpoint); the agent itself is the reasoning LLM. Trigger on: "remember this", "what have I learned about X", "should I do this / check this decision", "record the outcome", "what do I believe about X", "consolidate the brain", "set the objective", or when wiring an autonomous loop (a desk, a project) to durable, auditable memory. Supports MULTIPLE brains, each its own git repo.

Give an agent a persistent, evidence-first "brain" — a git repo it records experiences into and recalls grounded, cite-or-abstain answers from, with a deterministic constraint shield that vetoes "profitable but ruinous" decisions. Backed by the standalone `brain` CLI (Go, no network, no model endpoint); the agent itself is the reasoning LLM. Trigger on: "remember this", "what have I learned about X", "should I do this / check this decision", "record the outcome", "what do I believe about X", "consolidate the brain", "set the objective", or when wiring an autonomous loop (a desk, a project) to durable, auditable memory. Supports MULTIPLE brains, each its own git repo.