forgeplan_get

Returns a single artifact’s full markdown body plus frontmatter metadata. This is the canonical “read” operation — the agent calls it when it needs the actual contents of a PRD/RFC/ADR to reason about, quote, or update. Unlike forgeplan_list, the response contains everything the agent needs to understand the artifact.

Category: Reading Artifacts

When an agent calls this

User asks “what does PRD-042 say about rate limits?” — agent fetches then quotes the relevant section.
Before calling forgeplan_update — the agent needs the current body to produce a diff-aware patch.
Before forgeplan_reason — agent pre-reads so it can describe the context to the user if the LLM call is slow.

Input parameters

Name	Type	Required	Description
`id`	`string`	yes	Artifact ID to read (e.g. `PRD-042`, case-insensitive).

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

Returns

The artifact as a JSON object: all frontmatter fields (kind, status, depth, tags, dates, valid_until) plus the markdown body as a string. If the ID doesn’t exist, returns an error so the agent can recover (often by calling forgeplan_search to find the correct ID).

Example response shape:

{
  "id": "PRD-042",
  "kind": "prd",
  "status": "active",
  "depth": "standard",
  "title": "Authentication system",
  "tags": ["auth", "security"],
  "updated_at": "2026-04-11T09:31:00Z",
  "body": "# PRD-042: Authentication system\n\n## Problem\nUsers currently..."
}

Example invocation

{ "id": "PRD-001" }

With typical agent context:

User asks “remind me what PRD-042 says about token expiry”. Agent fetches the full body to quote the relevant section.

{ "id": "PRD-042" }

Typical sequence

forgeplan_list or forgeplan_search returns an ID → forgeplan_get pulls the body → agent quotes / reasons / updates. For bulk reads, agents should prefer forgeplan_search which returns ranked snippets rather than hitting get in a loop.

CLI equivalent

forgeplan show — human-readable terminal render