fix(runner): kill stale semantic-catalog counts and drift-check the doc

The catalog doc claimed 36 tools / 22 optional while the catalog holds 37 / 23
(the 36-vs-37 drift the reconciliation plan flagged). Correct the prose and add
src/catalog/catalog-docs.test.ts, which recomputes the total/always/optional
counts and the operation-id membership from the catalog so a stale hand count
can no longer pass.

Refs PAP-17025.

Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>

packages/paperclip-runner/docs/capability-semantic-tools.md

@@ -17,9 +17,12 @@ list, not a disposition — they are never exposed as tools. See
 [capability disposition](capability-disposition.md) and
 [authorization and exposure](capability-authorization-and-exposure.md).
 
-The catalog holds **36 tools**: 14 always-agent tools and 22 optional tools
+The catalog holds **37 tools**: 14 always-agent tools and 23 optional tools
 across 10 groups.
 
+<!-- These counts are drift-checked against src/tools/capability-semantic-tool-catalog.ts
+by src/catalog/catalog-docs.test.ts; update the catalog, not the numbers. -->
+
 ### Always-agent tools (14)
 
 `get_task_context`, `get_task_history`, `list_documents`, `read_document`,
@@ -27,7 +30,7 @@ across 10 groups.
 `finish_task`, `block_task`, `request_review`, `write_document`,
 `request_human_input`, `register_deliverable`, `inspect_operation_result`.
 
-### Optional tools (22), by group
+### Optional tools (23), by group
 
 | Group | Tools |
 | --- | --- |

packages/paperclip-runner/src/catalog/catalog-docs.test.ts

@@ -0,0 +1,60 @@
+import { readFileSync } from "node:fs";
+import { fileURLToPath } from "node:url";
+import { dirname, resolve } from "node:path";
+import { describe, expect, it } from "vitest";
+
+import {
+  CAPABILITY_SEMANTIC_TOOL_CATALOG as SCENARIO_CATALOG,
+  capabilitySemanticTool,
+} from "../tools/capability-semantic-tool-catalog.js";
+
+const docPath = resolve(
+  dirname(fileURLToPath(import.meta.url)),
+  "..",
+  "..",
+  "docs",
+  "capability-semantic-tools.md",
+);
+const doc = readFileSync(docPath, "utf8");
+
+/** The catalog counts region, between the disposition prose and descriptor shape. */
+function countsRegion(): string {
+  const start = doc.indexOf("## Two dispositions");
+  const end = doc.indexOf("## Descriptor shape");
+  expect(start).toBeGreaterThanOrEqual(0);
+  expect(end).toBeGreaterThan(start);
+  return doc.slice(start, end);
+}
+
+describe("capability-semantic-tools.md stays in sync with the catalog", () => {
+  const always = SCENARIO_CATALOG.filter((tool) => tool.disposition === "always_agent_tool");
+  const optional = SCENARIO_CATALOG.filter((tool) => tool.disposition === "optional_agent_tool");
+
+  it("states the current total, always, and optional counts", () => {
+    const region = countsRegion();
+    const total = /holds \*\*(\d+) tools\*\*/.exec(region)?.[1];
+    const alwaysProse = /(\d+) always-agent tools and (\d+) optional tools/.exec(region);
+    const alwaysHeader = /### Always-agent tools \((\d+)\)/.exec(region)?.[1];
+    const optionalHeader = /### Optional tools \((\d+)\)/.exec(region)?.[1];
+
+    expect(Number(total)).toBe(SCENARIO_CATALOG.length);
+    expect(Number(alwaysProse?.[1])).toBe(always.length);
+    expect(Number(alwaysProse?.[2])).toBe(optional.length);
+    expect(Number(alwaysHeader)).toBe(always.length);
+    expect(Number(optionalHeader)).toBe(optional.length);
+    // Guard against the historical 36-vs-37 drift the reconciliation plan flagged.
+    expect(SCENARIO_CATALOG.length).toBe(37);
+  });
+
+  it("lists exactly the catalog's operation ids in the counts region", () => {
+    const region = countsRegion();
+    const backticked = new Set(
+      [...region.matchAll(/`([a-z_]+)`/g)]
+        .map((match) => match[1])
+        .filter((id) => capabilitySemanticTool(id) !== undefined),
+    );
+    expect([...backticked].sort()).toEqual(
+      SCENARIO_CATALOG.map((tool) => tool.operationId).sort(),
+    );
+  });
+});