Feature Request

Define a standardized review request bundle format for agent-to-agent code/content reviews.

Problem

Today review packets are often assembled ad hoc (different files, fields, structure, naming). This creates avoidable friction:

• reviewers miss context,
• results are inconsistent,
• automation is harder,
• acceptance criteria are unclear.

Scope (important)

This request is about format/spec only.

It is not about transport/relay mechanism, mounts, or how bundles are physically delivered between agents.

Proposed Direction

Introduce a canonical review bundle schema (v1), e.g.:

• MANIFEST (metadata + version),
• REQUEST (goal, review type, constraints, priority),
• CONTEXT (relevant files/sections, dependencies),
• EXPECTATIONS (what output structure reviewer should return),
• ARTIFACTS list (paths, hashes, optional size/type),
• POLICY flags (sensitive data handling, redactions, allowed assumptions).

Why it matters

A stable format would:

• improve review quality and reproducibility,
• make reviewer behavior more predictable,
• enable tooling/validation around packet completeness,
• reduce prompt drift and accidental omission of critical context.

Related

• Discussion proposal (format-focused): will be linked
• File transfer in spawned agents (adjacent, not same scope):
  • PR sessions_spawn: inline attachments + redaction + hardening #16761
  • Issue [Feature]: sessions_spawn general file attachments (images/documents/archives) #17214

Ask

Agree on a minimal v1 schema and required fields so teams can start producing compatible review bundles.