Return compact search_agents results by default

CHANGELOG.md

@@ -13,6 +13,12 @@ parallel copies under `docs/` or `scripts/notes/`. At cut time: rename
 
 ## [Unreleased]
 
+### Changed
+
+- `search_agents` default results are id, description, and spawn metadata.
+  Pass `include_body=true` to include the loaded system prompt / body
+  (still truncated).
+
 ### Fixed
 
 - Dry-fleet transcript and `/status` report the outcome tally only
@@ -24,6 +30,7 @@ parallel copies under `docs/` or `scripts/notes/`. At cut time: rename
   `buzzing`, `grinding`, `thinking`, `doing`, `cooking`, `creating`,
   `imagining`, `inventing`) instead of going blank.
 
+
 ## [0.3.19] - 2026-09-10
 
 ### Security

docs/ARCHITECTURE.md

@@ -216,7 +216,7 @@ Workers ask the spawning parent with **`ask_director`** (not the human). That pa
 
 When the parent TUI is not blocked in `wait_agents`, the runner publishes an authoritative snapshot of currently pending top-level questions on each store notification, including empty snapshots before fleet-count updates. During synchronous session rotation, a runner-owned barrier suppresses both publications before delivery-generation invalidation, transcript clearing, and worker cancellation; successful reset reconciles a fresh snapshot before resuming asynchronous backend rebuild. The bridge drops resolved, cancelled, replaced, terminal, and removed asks and delivers each session/question identity once while pending. A coalesced wake starts only when the parent is not processing and every operator gate is closed, including parent-idle fleet holds where the shell stays busy. Worker gates do not manufacture parent processing. Replies use `send_input`'s `target` field with the worker session ID, never its shared catalog ID. Synthetic wakes use `SessionPort.deliver` through queued-delivery's idle-send path without entering the user follow-up queue or composer `/feedback` capture.
 
-When profiles exist (local `.agents/agents/` and/or enabled **`kind: "agent"`** plugins, including **data-only** markdown plugins with no `index.ts`), the chat model also receives **`search_agents`** — a lexical index over profile id, description, and role text so the model can discover ids before calling `spawn_agent(agent=...)`. Results include each match's full loaded system prompt / body so the parent can inspect plugin or Claude marketplace agents without `read_file` on paths outside the session cwd (path-escape blocks those roots by design; writes remain blocked). `spawn_agent` and `search_agents` are core tools on the primary session.
+When profiles exist (local `.agents/agents/` and/or enabled **`kind: "agent"`** plugins, including **data-only** markdown plugins with no `index.ts`), the chat model also receives **`search_agents`** — a lexical index over profile id, description, and role text so the model can discover ids before calling `spawn_agent(agent=...)`. Default results are id, description, and spawn metadata (orchestrator flag, source). Pass `include_body=true` to include each match's loaded system prompt / body (truncated) so the parent can inspect plugin or Claude marketplace agents without `read_file` on paths outside the session cwd (path-escape blocks those roots by design; writes remain blocked). `spawn_agent` and `search_agents` are core tools on the primary session.
 
 Built-in directors with `spawn.maySpawn` may themselves call `spawn_agent` (one hop only): nested dispatch installs the mailbox-scoped fleet verbs (`spawn_agent`, `wait_agents`, `list_agents`, …) with `allowOrchestrator: false` so the tree bottoms out. Profile-sourced `orchestrator: true` is rejected before a session starts because it has no trusted tier/authority semantics today. Fleet discovery (`search_agents`) stays Tier 1 only. Unknown `agent` ids fail closed.
 

docs/PLUGINS.md

@@ -170,8 +170,9 @@ Relative install paths and paths outside that root are ignored so a poisoned
 registry cannot load project trees as origin `user`. Profile `source: "claude"`.
 Discovered modules still require `settings.plugins[id].enabled` before agents or
 tools wire into the session. `search_agents` labels those profiles with
-`[source: claude]` and injects each profile's full loaded system prompt / body so
-the parent never needs `read_file` on `~/.claude/plugins/...` (path-escape still
+`[source: claude]`. Default results are id, description, and spawn metadata;
+pass `include_body=true` to include the loaded system prompt / body so the
+parent never needs `read_file` on `~/.claude/plugins/...` (path-escape still
 blocks those roots for path tools; writes/deletes outside cwd stay denied). JS
 Claude plugins (if any) stay on explicit `pluginPaths`.
 

src/agent/agent-search.test.ts

@@ -45,60 +45,92 @@ describe("createAgentIndex", () => {
 
 describe("formatAgentSearchResults", () => {
   test("includes spawn hint and ids", () => {
-    const text = formatAgentSearchResults([defined(fixtures[1])]);
+    const text = formatAgentSearchResults([defined(fixtures[1])], false);
     expect(text).toContain("critique");
     expect(text).toContain("spawn_agent(agent=");
   });
 
   test("includes source label when present", () => {
-    const text = formatAgentSearchResults([
-      {
-        id: "marketplace-scout",
-        description: "From Claude install",
-        source: "claude",
-      },
-    ]);
+    const text = formatAgentSearchResults(
+      [
+        {
+          id: "marketplace-scout",
+          description: "From Claude install",
+          source: "claude",
+        },
+      ],
+      false,
+    );
     expect(text).toContain("[source: claude]");
     expect(text).toContain("marketplace-scout");
   });
 
-  test("injects full system prompt body so parent need not read_file plugin roots", () => {
+  test("default omits system prompt body", () => {
+    const uniqueBody =
+      "UNIQUE_BODY_MARKER_draper_review_pull_requests_for_design_clarity_xyz";
+    const text = formatAgentSearchResults(
+      [
+        {
+          id: "draper",
+          description: "PR design reviewer from marketplace",
+          source: "claude",
+          orchestrator: true,
+          systemPromptRole: uniqueBody,
+        },
+      ],
+      false,
+    );
+    expect(text).toContain("### draper");
+    expect(text).toContain("[source: claude]");
+    expect(text).toContain("[orchestrator]");
+    expect(text).toContain("PR design reviewer from marketplace");
+    expect(text).not.toContain(uniqueBody);
+    expect(text).not.toContain("System prompt / body:");
+  });
+
+  test("include_body injects truncated system prompt body", () => {
     const body =
       "You are draper. Review pull requests for design clarity and maintainability.\n" +
       "Prefer concrete file/line citations.";
-    const text = formatAgentSearchResults([
-      {
-        id: "draper",
-        description: "PR design reviewer from marketplace",
-        source: "claude",
-        systemPromptRole: body,
-      },
-    ]);
+    const text = formatAgentSearchResults(
+      [
+        {
+          id: "draper",
+          description: "PR design reviewer from marketplace",
+          source: "claude",
+          systemPromptRole: body,
+        },
+      ],
+      true,
+    );
     expect(text).toContain("### draper");
     expect(text).toContain("[source: claude]");
     expect(text).toContain("System prompt / body:");
     expect(text).toContain(body);
-    expect(text).toContain("do not need read_file on plugin roots");
   });
 
   test("omits body section when systemPromptRole is absent", () => {
-    const text = formatAgentSearchResults([
-      { id: "no-body", description: "Metadata only" },
-    ]);
+    const text = formatAgentSearchResults(
+      [{ id: "no-body", description: "Metadata only" }],
+      true,
+    );
     expect(text).toContain("### no-body");
     expect(text).toContain("Metadata only");
     expect(text).not.toContain("System prompt / body:");
   });
 
   test("truncates oversized systemPromptRole bodies with ellipsis marker", () => {
     const body = "x".repeat(MAX_AGENT_SEARCH_BODY_CHARS + 500);
-    const text = formatAgentSearchResults([
-      {
-        id: "huge",
-        description: "Oversized marketplace body",
-        systemPromptRole: body,
-      },
-    ]);
+    const text = formatAgentSearchResults(
+      [
+        {
+          id: "huge",
+          description: "Oversized marketplace body",
+          systemPromptRole: body,
+        },
+      ],
+      true,
+    );
     expect(text).toContain("System prompt / body:");
     expect(text).toContain("…[truncated]");
     expect(text).not.toContain(body);
@@ -113,14 +145,17 @@ describe("formatAgentSearchResults", () => {
   test("redacts secret-shaped content in profile body at format layer", () => {
     // search_agents is not on the posix middleware path; scrub must happen here.
     const secret = "sk-live-abc123xyz789012345678";
-    const text = formatAgentSearchResults([
-      {
-        id: "leaky",
-        description: "Profile with credential-shaped body text",
-        source: "claude",
-        systemPromptRole: `Use API_KEY=${secret} when calling the provider.`,
-      },
-    ]);
+    const text = formatAgentSearchResults(
+      [
+        {
+          id: "leaky",
+          description: "Profile with credential-shaped body text",
+          source: "claude",
+          systemPromptRole: `Use API_KEY=${secret} when calling the provider.`,
+        },
+      ],
+      true,
+    );
     expect(text).toContain("### leaky");
     expect(text).toContain("System prompt / body:");
     expect(text).toContain(CREDENTIAL_REDACTION);
@@ -130,7 +165,35 @@ describe("formatAgentSearchResults", () => {
 });
 
 describe("createSearchAgentsTool", () => {
-  test("handler surfaces loaded systemPromptRole for plugin-style profiles", async () => {
+  test("handler default omits systemPromptRole body", async () => {
+    const uniqueBody =
+      "UNIQUE_BODY_MARKER_emil_product_sense_and_user_impact_xyz";
+    const tool = createSearchAgentsTool(() => [
+      {
+        id: "emil",
+        description: "Product-minded reviewer",
+        source: "claude",
+        systemPromptRole: uniqueBody,
+      },
+      {
+        id: "greybeard",
+        description: "Architect",
+        systemPromptRole: "You are greybeard.",
+      },
+    ]);
+    if (tool.kind !== "string") throw new Error("expected string tool");
+    const text = await tool.handler(
+      { query: "emil product" },
+      new AbortController().signal,
+    );
+    expect(text).toContain("emil");
+    expect(text).toContain("Product-minded reviewer");
+    expect(text).toContain("[source: claude]");
+    expect(text).not.toContain(uniqueBody);
+    expect(text).not.toContain("System prompt / body:");
+  });
+
+  test("handler include_body true surfaces loaded systemPromptRole", async () => {
     const body = "You are emil. Focus on product sense and user impact.";
     const tool = createSearchAgentsTool(() => [
       {
@@ -147,7 +210,7 @@ describe("createSearchAgentsTool", () => {
     ]);
     if (tool.kind !== "string") throw new Error("expected string tool");
     const text = await tool.handler(
-      { query: "emil product" },
+      { query: "emil product", include_body: true },
       new AbortController().signal,
     );
     expect(text).toContain("emil");
@@ -199,7 +262,7 @@ describe("createSearchAgentsTool", () => {
     ]);
     if (tool.kind !== "string") throw new Error("expected string tool");
     const text = await tool.handler(
-      { query: "leaky" },
+      { query: "leaky", include_body: true },
       new AbortController().signal,
     );
     expect(text).toContain("### leaky");

src/agent/agent-search.ts

@@ -72,38 +72,42 @@ function truncateAgentBody(body: string): string {
   return `${body.slice(0, MAX_AGENT_SEARCH_BODY_CHARS)}\n…[truncated]`;
 }
 
-// Format one profile for search_agents output. Injects the full loaded
-// systemPromptRole (markdown body / role text) so the parent can inspect plugin
-// and marketplace agents without read_file on paths outside the session cwd
-// (path-escape blocks those roots by design). Bodies longer than
+// Format one profile for search_agents output. Default is id, description, and
+// spawn metadata (orchestrator flag, source). The loaded systemPromptRole is
+// omitted unless includeBody is true. Bodies longer than
 // MAX_AGENT_SEARCH_BODY_CHARS are truncated with an ellipsis marker.
-function formatAgentProfileEntry(p: AgentProfile): string {
+function formatAgentProfileEntry(
+  p: AgentProfile,
+  includeBody: boolean,
+): string {
   const desc = (p.description ?? "").trim();
   const orch = p.orchestrator === true ? " [orchestrator]" : "";
   const source = p.source !== undefined ? ` [source: ${p.source}]` : "";
   const header =
     desc.length > 0
       ? `### ${p.id}${orch}${source}\n${desc}`
       : `### ${p.id}${orch}${source}`;
+  if (!includeBody) return header;
   const body = (p.systemPromptRole ?? "").trim();
   if (body.length === 0) return header;
   return `${header}\n\nSystem prompt / body:\n${truncateAgentBody(body)}`;
 }
 
 export function formatAgentSearchResults(
   profiles: readonly AgentProfile[],
+  includeBody: boolean,
 ): string {
   if (profiles.length === 0) {
     return "No agent profiles matched. Try broader terms (e.g. review, explore, implement) or list_dir on .agents/agents/.";
   }
-  const entries = profiles.map(formatAgentProfileEntry);
+  const entries = profiles.map((p) => formatAgentProfileEntry(p, includeBody));
   // Live scrub for search_agents: this tool is not on the posix middleware path, so
   // SCRUBBABLE_TOOLS in tool-result-secret-scrub-plugin cannot reach it. Scrub here
   // before the formatted string becomes a tool result (marketplace/plugin bodies may
   // contain secret-shaped substrings).
   return scrubSecretShapedContent(
     [
-      "Matching agent profiles (pass id to spawn_agent(agent=...)). Full system prompt / body is included so you do not need read_file on plugin roots outside the workspace:",
+      "Matching agent profiles (pass id to spawn_agent(agent=...)):",
       "",
       ...entries.flatMap((entry, i) => (i === 0 ? [entry] : ["", entry])),
       "",
@@ -115,7 +119,7 @@ export function formatAgentSearchResults(
 export const searchAgentsDefinition: ToolDefinition = {
   name: "search_agents",
   description:
-    "Find spawnable agent profiles by capability, role, or team name (e.g. 'review', 'review team', 'architect', 'security'). Returns profile ids, descriptions, and the full loaded system prompt / body for each match so you can inspect plugin or Claude marketplace agents without reading files outside the workspace. Use the id in spawn_agent(agent=...). Call this when the user asks to spin up specialists or a team without naming exact ids.",
+    "Find spawnable agent profiles by capability, role, or team name (e.g. 'review', 'review team', 'architect', 'security'). Returns profile ids, descriptions, and spawn metadata (orchestrator flag, source). Pass include_body=true to include the loaded system prompt / body for each match (truncated). Use the id in spawn_agent(agent=...). Call this when the user asks to spin up specialists or a team without naming exact ids.",
   inputSchema: {
     type: "object",
     properties: {
@@ -124,12 +128,17 @@ export const searchAgentsDefinition: ToolDefinition = {
         description:
           "What kind of agent or team you need — keywords from the user's request (e.g. 'review team', 'code quality', 'explore codebase').",
       },
+      include_body: {
+        type: "boolean",
+        description:
+          "When true, include each match's loaded system prompt / body (truncated). Default false: id, description, and spawn metadata only.",
+      },
     },
     required: ["query"],
   },
 };
 
-const SearchAgentsArgs = type({ query: "string" });
+const SearchAgentsArgs = type({ query: "string", "include_body?": "boolean" });
 
 export function createSearchAgentsTool(
   getProfiles: () => readonly AgentProfile[],
@@ -140,15 +149,16 @@ export function createSearchAgentsTool(
     handler: async (rawArgs: Record<string, unknown>): Promise<string> => {
       const parsed = SearchAgentsArgs(rawArgs);
       if (parsed instanceof type.errors) {
-        return "Error: search_agents requires query (string).";
+        return "Error: search_agents requires query (string); include_body is optional boolean.";
       }
       const query = parsed.query.trim();
+      const includeBody = parsed.include_body === true;
       // Empty and non-empty queries share createAgentIndex.search's default limit
       // (12) so a large marketplace catalog cannot dump every body into one result.
       if (query.length === 0 && getProfiles().length === 0) {
         return "No agent profiles are loaded.";
       }
-      return formatAgentSearchResults(index.search(query));
+      return formatAgentSearchResults(index.search(query), includeBody);
     },
   });
 }

src/agent/prompts.ts

@@ -93,7 +93,7 @@ export function buildHarnessFacts(
     ...(dynamicTools
       ? [
           "- Core tools plus the advertised catalog (including skill_search) are resident. Use tool_search to load extra capabilities from plugins or integrations when needed.",
-          "- Use search_agents before dispatching named specialists or teams (results include full profile bodies; do not read_file plugin paths outside the workspace).",
+          "- Use search_agents before dispatching named specialists or teams (ids and descriptions by default; include_body=true for the loaded system prompt). Do not read_file plugin paths outside the workspace.",
           "- The user may send follow-up messages while workers run; they are queued. Enter delivers at the next parent tool.boundary; Alt+Enter on session-idle. A long parent tool holds that boundary. Update your plan, spawn or adjust workers, and keep the operator informed.",
         ]
       : ["- The tools below are your full toolset."]),
@@ -245,7 +245,7 @@ const TOOL_SUMMARIES: Record<string, string> = {
   wait_agents:
     "wait for spawned workers by agent_id; returns awaiting_director when a worker asks, without collecting that session",
   search_agents:
-    "find agent profiles by role or team before spawning with spawn_agent(agent=...); results include full system prompt / body so you need not read_file plugin roots outside the workspace",
+    "find agent profiles by role or team before spawning with spawn_agent(agent=...); default results are id, description, and spawn metadata — pass include_body=true for the loaded system prompt / body",
   manage_tasks:
     "maintain your work checklist — create/replace, update status, append, cancel",
   ask_director:

src/subagent/agent-fleet.ts

@@ -768,8 +768,8 @@ function resolveAgentDispatch(input: {
       const known = profiles.map((p) => p.id).sort();
       const hint =
         known.length > 0
-          ? ` Known profiles: ${known.join(", ")}. Call search_agents to discover more (results include full system prompt / body; do not read_file plugin paths outside the workspace).`
-          : " No profiles are currently loaded. Call search_agents to discover available agents (results include full system prompt / body).";
+          ? ` Known profiles: ${known.join(", ")}. Call search_agents to discover more.`
+          : " No profiles are currently loaded. Call search_agents to discover available agents.";
       return { error: `Error: unknown agent profile "${agentId}".${hint}` };
     }
     let effortPin: ReasoningEffort | undefined;