forgeplan_new

Creates a new artifact stub from the built-in template for its kind. The agent calls this when it has decided (usually via forgeplan_route) what kind of artifact is needed next — typically the moment after the human request is classified as Standard or deeper. The returned ID is the handle the agent uses for every subsequent operation.

Category: Creating Artifacts

When an agent calls this

After forgeplan_route returns Depth: Standard, Pipeline: PRD → RFC and the agent needs the PRD stub.
When the agent is told to “create an ADR for X decision we just made” and wants a skeleton to fill in.
When decomposing work — after forgeplan_decompose suggests RFCs, the agent may call new once per suggested RFC.

Input parameters

Name	Type	Required	Description
`kind`	`string`	yes	Artifact kind: `prd`, `epic`, `spec`, `rfc`, `adr`, `problem`, `solution`, `evidence`, `note`, `refresh`.
`title`	`string`	yes	Artifact title.

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

Returns

The generated ID plus the rendered body. The agent should immediately fill MUST sections via forgeplan_update rather than leave a stub — unpopulated PRDs count as blind spots in forgeplan_health.

Example response shape:

{
  "id": "PRD-042",
  "kind": "prd",
  "status": "draft",
  "path": ".forgeplan/prds/prd-042-authentication-system.md",
  "body": "# PRD-042: Authentication system\n\n## Problem\n..."
}

Example invocation

{ "kind": "prd", "title": "Authentication system" }

With typical agent context:

Router returned Depth: Standard, Pipeline: PRD → RFC. Agent creates the PRD stub before touching code.

{ "kind": "prd", "title": "Rate limit auth endpoints" }

Typical sequence

forgeplan_route → forgeplan_new → forgeplan_update (fill MUST sections) → forgeplan_validate (PASS) → forgeplan_reason (ADI for Standard+) → code → evidence → forgeplan_activate.

CLI equivalent

forgeplan new — same operation with interactive prompts