The workspace README (one-read orientation)

Read workspace_readme://current to get a live map of your workspace's pages, knowledge, workflows, and callable operations in a single call

The workspace README is a live orientation map of one workspace, assembled fresh on every read. It lists the workspace's pages, knowledge documents, agent workflows, and callable operation namespaces as name-plus-summary rows, and includes the same content rendered as one markdown document. Nothing is stored, so it can never go stale.

Why it exists

An agent connecting to a workspace normally makes four discovery calls (pages, knowledge base, workflows, operations) before it can do anything useful, and usually guesses an id or slug wrong along the way. The README replaces that with one read. Point your agent at it first:

MCP: read the resource workspace_readme://current
REST: GET https://app.amdahl.co/api/platform/v1/workspace/readme

What comes back

json

{
  "workspace_slug": "acme",
  "generated_at": "2026-07-07T12:00:00.000Z",
  "sections": [
    {
      "key": "pages",
      "title": "Pages",
      "source_uri": "page://list",
      "count": 3,
      "items": [
        {
          "name": "Pipeline Health",
          "summary": "Live deal-stage dashboard",
          "uri": "page://8c0e…"
        }
      ],
      "omitted": false,
      "omitted_reason": null
    }
  ],
  "markdown": "# Workspace README — acme\n…"
}

Four sections, in order:

Section	Source	What each row is
Pages	`page://list`	A saved data page you can open or render
Knowledge	`knowledge_base://list`	A reference document (current version per family)
Agents & Workflows	`agent_blueprint://list`	A workflow recipe; badges mark `starter` and `active`
Actions	`operation://list`	One row per operation namespace, with its operation ids

Every row carries the resource uri that fetches the full object, so the README doubles as a jump table: read it, pick a row, follow the URI.

The markdown field is the same content as one shareable document. Copy it into a note, paste it into an onboarding doc, or hand it to a teammate.

Permissions

The read itself requires only artifacts:read, so every key tier, including read-only keys, can call it. Each section additionally honors its own scope (pages:read, knowledge_base:read, and so on). A section your key cannot read comes back with omitted: true and a reason instead of failing the whole call, and the same note appears in the markdown, so the document shape is stable across key tiers.

Practical notes

Section row counts follow each source's default page size; count reports the source's total, and the markdown marks any overflow with a pointer back to the source URI.
The Actions section is deliberately a namespace map, not the full catalog. Drill into one namespace via operation://list?namespace=<name>, or read the tool catalog for every description.
Reading the README costs a handful of list queries per call. If you are polling, cache it client-side; it is meant to be read once per session, not on a loop.