@dharma-ai-labs/agent-fabric@0.2.68
Local-first CLI for connecting coding agents to Dharma evaluation, orchestration, and signed Skill delivery
AI security review
scanned · by lpm-firewall-aiSource shows package-owned agent setup that can alter Claude project permissions, making this a warning-level agent capability. The manifest has no npm install hook, and the rules are described in code as bounded read-only permissions.