Review System for Enterprise & Agent Mail Use Cases

[kraitsura]

I'd like to propose adding a simple review/approval system to Beads for enterprise environments where human oversight of AI-generated plans and tasks is required before execution.

Motivation:

• Optionally keeps human in the loop
• Allows Orgs to use this as a s serious oversight and collaboration tool
• Allows for fine-grained feedback loop for the LLM
  • make reviews and ask agent to fix plans for epic,
  • agents talk within themselves to review and approve plans if needed

Proposed Changes:

• New reviews table & synced JSONL file
• approval_status field on issues
• Give MCP read write permissions

Backwards compatible: Review is opt-in; existing workflows unchanged

Wanted to gauge interest in taking the tool in this direction and whether it falls in line with the core workflow and design philosophy of beads. I personally think it can add a ton of value with very little overhead.

[kraitsura]

I have a PR ready if interested.

[mdlmarkham]

I like the idea