forgeplan_review

Review a single artifact — runs validation (depth-aware MUST / SHOULD rules), checks lifecycle prerequisites (evidence, R_eff, linked relations), and returns a clear verdict on whether the artifact is ready to activate. This is the combined “can I activate?” gate used before calling forgeplan_activate.

Category: Quality

When an agent calls it

Before activation — confirm all gates pass to avoid activation errors.
PR review — run on each touched artifact to catch missing sections or MUST failures.
Author self-check — faster than running validate + score + manual lifecycle inspection.
Automated quality hooks — pre-commit / CI can call this and fail the build if any artifact regresses.

Input parameters

Name	Type	Required	Description
`id`	`string`	yes	Artifact ID to review.

Schema source: crates/forgeplan-mcp/src/server.rs::ReviewParams

Returns

{
  "artifact_id": "PRD-042",
  "kind": "prd",
  "depth": "standard",
  "status": "draft",
  "validation": {
    "must_errors": [],
    "should_warnings": ["density < 50 words in section Goals"]
  },
  "lifecycle": {
    "r_eff": 0.72,
    "has_evidence": true,
    "ready_to_activate": true
  },
  "verdict": "PASS — ready to activate"
}

If blocked:

{
  "verdict": "FAIL",
  "validation": { "must_errors": ["Missing section: Problem"] },
  "lifecycle": { "ready_to_activate": false }
}

Example invocation

{ "id": "PRD-042" }

Typical sequence

forgeplan_review → if FAIL, fix issues.
forgeplan_update to patch the body.
forgeplan_review again → expect PASS.
forgeplan_activate — flip draft → active.

CLI equivalent

forgeplan review PRD-042