@llamaventures/cli 1.15.1 → 1.16.0

package/AGENT_BRIEFING.md

@@ -1,6 +1,6 @@
 # Llama Ventures Agent Briefing
 
-You've been onboarded as a teammate of [Llama Ventures](https://llamaventures.vc) via the `@llamaventures/cli` package. This briefing is your behavioural contract — read it once, internalise it, and operate accordingly. The user shouldn't have to explain any of this to you again.
+You've been onboarded as a teammate of [Llama Ventures](https://llamaventures.vc) via the `@llamaventures/cli` package. This bundled briefing is a fallback copy of the behavioural contract. When authenticated, prefer the server-owned contract from `llama agent-onboard`, `llama agent bootstrap`, or MCP `agent_briefing`; read it once, internalise it, and operate accordingly. The user shouldn't have to explain any of this to you again.
 
 You are not just an AI assistant. You're an **extension of a team member** — with CLI access to the Llama Command pipeline, write permission on shared data, and audit-log responsibility. Treat the status seriously.
 
@@ -25,6 +25,7 @@ Most teammates don't know everything this CLI can do. Part of your job is to sur
 This npm package is public, but Llama OS skills are private. Do not assume the skill text is bundled locally. For team-token sessions, discover the live runtime library through Llama Command:
 
 - Start with `llama agent bootstrap` or MCP `agent_bootstrap` when you need the current Command + Llama OS contract.
+- Use `llama agent-onboard` or MCP `agent_briefing` for the server-owned Agent Runtime Contract. The npm-bundled text is fallback only.
 - Use `llama skills search "<task>"` or MCP `skills_search` before choosing a Llama workflow.
 - Use `llama skills show <slug>` or MCP `skills_read` only for the relevant skill.
 - Use `llama explain <command-url-or-object>` or MCP `object_inspect` for 404s, deleted wiki pages, notifier links, deal URLs, and unknown Command objects before telling the user "the system is broken."
@@ -337,9 +338,9 @@ the intake agent; relay the conversation, then surface the verdict.
 
 ## Where this content lives
 
-- This file: `AGENT_BRIEFING.md` in the `@llamaventures/cli` npm package
-- Re-printable any time: `llama agent-onboard`
-- MCP prompt: `agent_briefing` from the `llama-mcp` server
+- Canonical when authenticated: `GET /api/agent/briefing` via `llama agent-onboard` or MCP `agent_briefing`
+- Runtime home screen: `GET /api/agent/manifest` via `llama agent bootstrap` or MCP `agent_bootstrap`
+- Fallback copy: `AGENT_BRIEFING.md` in the `@llamaventures/cli` npm package
 - Public source: <https://github.com/Llama-Ventures/llama-cli/blob/main/AGENT_BRIEFING.md>
 
 You're now a Llama Ventures teammate (or an external agent helping a

package/CHANGELOG.md

@@ -6,6 +6,27 @@ this project adheres to [Semantic Versioning](https://semver.org).
 
 ## [Unreleased]
 
+## [1.16.0] - 2026-06-22
+
+### Added
+- CLI and MCP requests now send bounded local-agent telemetry metadata to
+  Llama Command: client kind/version, detected agent client, session id,
+  normalized command, latency, status, sanitized args, result ids, and bounded
+  result summaries.
+- `llama eval good|bad --last` and `llama eval add "<query>" --expect ...`
+  let agents and humans turn real search usage into Golden Query Eval
+  feedback.
+- MCP tool `record_eval_feedback` mirrors the CLI eval feedback path for
+  MCP-native agents.
+
+### Changed
+- `llama agent-onboard` and MCP `agent_briefing` now prefer the authenticated
+  Llama Command `/api/agent/briefing` runtime contract, with bundled
+  `AGENT_BRIEFING.md` retained as fallback only.
+- `llama agent bootstrap` and MCP `agent_bootstrap` now pass the local CLI
+  version to Command so the server can return stale/ok CLI guidance as part of
+  the agent contract.
+
 ## [1.15.1] — 2026-06-16
 
 ### Changed

package/README.md

@@ -226,6 +226,8 @@ llama agent bootstrap
 llama skills search "wiki delete tombstone"
 llama skills show llama-command
 llama explain https://command.llamaventures.vc/wiki/some-page
+llama eval bad --last --reason "missed the llamaos weekly note"
+llama eval add "last week llama dev weekly" --expect wiki:llamaos-weekly-2026-06-17
 
 # Wiki
 llama wiki search "<query>"
@@ -262,10 +264,31 @@ agents can pattern-match without parsing prose.
 
 ---
 
+### Golden Query Eval feedback
+
+The CLI and MCP server send lightweight client telemetry to Command for each
+authenticated request: client kind/version, detected agent client, local session
+id, normalized command, sanitized args, canonical result ids, status, latency,
+and bounded summaries. It records what Llama Command was asked to do, not the
+user's private Claude Code/Codex/Cursor conversation or local files.
+
+Search commands automatically become eval candidates. Agents can mark the latest
+result explicitly:
+
+```bash
+llama eval good --last
+llama eval bad --last --reason "wrong top wiki"
+llama eval add "AI陪伴" --surface deal --expect deal:<uuid>
+```
+
+MCP-native agents use `record_eval_feedback` for the same flow.
+
+---
+
 ## MCP server
 
 The bundled `llama-mcp` is a **stdio Model Context Protocol** server exposing
-**56 typed tools** that mirror the most-used CLI surface. Every tool is named
+typed tools that mirror the most-used CLI surface. Every tool is named
 and scoped — there is no generic API passthrough, by design (a public-package
 escape hatch reachable from a prompt-injectable agent context is exactly the
 shape we want to avoid).
@@ -290,9 +313,11 @@ printf '%s\n' \
 ```
 
 Auth is identical to the CLI's chain (gcloud → `$LLAMA_TOKEN` → `~/.llama/token`).
-The `agent_briefing` MCP **prompt** also returns
-[`AGENT_BRIEFING.md`](AGENT_BRIEFING.md) verbatim, so any new agent loading the
-server can self-onboard without leaving the protocol.
+The `agent_briefing` MCP **prompt** returns the server-owned Agent Runtime
+Contract when authenticated, so any new agent loading the server can
+self-onboard without leaving the protocol. The bundled
+[`AGENT_BRIEFING.md`](AGENT_BRIEFING.md) is only a fallback if the server
+briefing route is temporarily unavailable.
 
 For current Llama OS skills, use the runtime tools instead of looking for a
 local private repo: `agent_bootstrap`, `skills_search`, `skills_read`, and
@@ -352,8 +377,8 @@ extensions, no transport flags.
 
 > If you're new and want the agent to onboard itself, run
 > `llama agent-onboard` from the CLI or fetch the `agent_briefing` prompt from
-> the MCP server. It's the workflow contract — autonomy levels, attribution
-> grammar, error recovery, anti-pollution rules.
+> the MCP server. It pulls the Command-owned workflow contract — current CLI
+> guidance, attribution grammar, error recovery, and anti-pollution rules.
 
 ---
 

package/README.zh-CN.md

@@ -299,9 +299,10 @@ printf '%s\n' \
 ```
 
 认证链跟 CLI 完全一样（gcloud → `$LLAMA_TOKEN` → `~/.llama/token`）。
-`agent_briefing` 这个 MCP **prompt** 还会原样吐出
-[`AGENT_BRIEFING.md`](AGENT_BRIEFING.md)——刚装上 server 的 agent
-不用离开协议就能给自己 onboard。
+`agent_briefing` 这个 MCP **prompt** 会在认证后拉取 Command 服务端的
+Agent Runtime Contract——刚装上 server 的 agent 不用离开协议就能给自己
+onboard。包内 [`AGENT_BRIEFING.md`](AGENT_BRIEFING.md) 只是服务端 briefing
+临时不可用时的 fallback。
 
 要读当前 Llama OS skills，用 runtime tools：`agent_bootstrap`、
 `skills_search`、`skills_read`、`object_inspect`。公开 npm 包不打包私有
@@ -358,7 +359,7 @@ claude mcp add llama -- llama-mcp
 </details>
 
 > 想让 agent 自己 onboard？跑 `llama agent-onboard`，或者从 MCP server
-> 拉 `agent_briefing` prompt——那是 AI agent 的工作合约（autonomy 等级、
+> 拉 `agent_briefing` prompt——它会拿 Command 服务端的工作合约（CLI 状态、
 > attribution 语法、错误恢复、anti-pollution 规则）。
 
 ---

package/bin/llama-mcp.mjs

@@ -11,7 +11,14 @@ import { createRequire } from "module";
 import { McpServer } from "@modelcontextprotocol/sdk/server/mcp.js";
 import { StdioServerTransport } from "@modelcontextprotocol/sdk/server/stdio.js";
 import { z } from "zod";
-import { getAuthHeaders, readBriefing, request, requestSse } from "../lib/client.mjs";
+import {
+  getAuthHeaders,
+  getLastAgentEvent,
+  readBriefing,
+  request,
+  requestSse,
+  setClientRuntime,
+} from "../lib/client.mjs";
 
 const requireFromHere = createRequire(import.meta.url);
 const { version: PKG_VERSION } = requireFromHere("../package.json");
@@ -23,6 +30,8 @@ import {
   uploadExternalFile,
 } from "../lib/external.mjs";
 
+setClientRuntime({ client: "mcp" });
+
 // Wrap a request() call into the MCP CallToolResult shape. Catches errors
 // (NO_AUTH / 401 / 5xx / network) and surfaces them as `isError: true`
 // content so the calling agent sees a clean error string instead of the
@@ -56,6 +65,20 @@ function splitSources(value) {
     .filter(Boolean);
 }
 
+function expectedIds(value) {
+  const expected = { dealIds: [], wikiSlugs: [], raw: [] };
+  if (!value) return expected;
+  const items = Array.isArray(value) ? value : String(value).split(",");
+  for (const item of items.map((s) => String(s).trim()).filter(Boolean)) {
+    const [kind, ...rest] = item.split(":");
+    const id = rest.join(":").trim();
+    if (kind === "deal" && id) expected.dealIds.push(id);
+    else if ((kind === "wiki" || kind === "slug") && id) expected.wikiSlugs.push(id);
+    else expected.raw.push(item);
+  }
+  return expected;
+}
+
 function buildEnrichmentAgentMessage(args = {}) {
   if (args.message) return String(args.message);
   const sources = splitSources(args.sources) ?? [
@@ -209,11 +232,47 @@ server.registerTool(
   },
   async ({ limit } = {}) => {
     const params = new URLSearchParams();
+    params.set("clientVersion", PKG_VERSION);
     if (limit) params.set("limit", String(limit));
     return callApi("GET", `/api/agent/manifest${params.toString() ? `?${params}` : ""}`);
   }
 );
 
+server.registerTool(
+  "record_eval_feedback",
+  {
+    description:
+      "Mark the latest llama CLI/MCP result as good/bad or add a real query " +
+      "to the Golden Query Eval candidate pool. Use when the user says a " +
+      "Llama Command search/result was right, wrong, missing a source, or should be regression-tested.",
+    inputSchema: {
+      action: z.enum(["good", "bad", "add"]).describe("feedback action"),
+      eventId: z.number().optional().describe("agent_client_events id; defaults to latest local event"),
+      query: z.string().optional().describe("required for manual add when no source event is available"),
+      surface: z.string().optional().describe("deal, wiki, activity, people, or manual"),
+      expect: z
+        .union([z.string(), z.array(z.string())])
+        .optional()
+        .describe("expected ids like wiki:llamaos-weekly-2026-06-17 or deal:<uuid>"),
+      reason: z.string().optional().describe("why this was good/bad or should be tracked"),
+      privacyLevel: z.string().optional().describe("default internal"),
+    },
+  },
+  async ({ action, eventId, query, surface, expect, reason, privacyLevel }) => {
+    const last = getLastAgentEvent();
+    const body = {
+      action,
+      eventId: eventId ?? last?.lastEventId,
+      query,
+      surface: surface ?? last?.lastSurface,
+      expected: expectedIds(expect),
+      reason,
+      privacyLevel: privacyLevel ?? "internal",
+    };
+    return callApi("POST", "/api/agent/eval-feedback", body);
+  }
+);
+
 server.registerTool(
   "skills_search",
   {
@@ -1595,11 +1654,12 @@ server.registerPrompt(
       "skills_search, and skills_read.",
   },
   async () => {
-    // Gate the briefing behind a /api/me check so unauthenticated MCP
-    // clients can't harvest internal workflow / command surface just by
-    // requesting the prompt. Mirrors the CLI gate in bin/llama.mjs.
+    // Gate the briefing behind authenticated Command runtime. The server-owned
+    // /api/agent/briefing contract is canonical; bundled AGENT_BRIEFING.md is
+    // only a rollout/offline fallback for authenticated users.
     const headers = await getAuthHeaders();
     let stub = null;
+    let briefing = null;
     if (Object.keys(headers).length === 0) {
       stub =
         "Llama Ventures team onboarding requires credentials.\n\n" +
@@ -1610,19 +1670,28 @@ server.registerPrompt(
         "Founder / external visitor: use the `pitch_*` tools — no token required.";
     } else {
       try {
-        await request("GET", "/api/me");
+        const params = new URLSearchParams({ clientVersion: PKG_VERSION });
+        const body = await request("GET", `/api/agent/briefing?${params}`);
+        briefing = body?.briefing || null;
       } catch (err) {
-        stub =
-          "Llama Ventures team onboarding requires valid credentials. " +
-          "Server rejected the credentials we sent. Re-mint at " +
-          "https://command.llamaventures.vc/settings/tokens.";
+        const msg = err?.message || "";
+        if (msg.includes("Error[UNAUTHORIZED]") || msg.includes("Error[NO_AUTH]")) {
+          stub =
+            "Llama Ventures team onboarding requires valid credentials. " +
+            "Server rejected the credentials we sent. Re-mint at " +
+            "https://command.llamaventures.vc/settings/tokens.";
+        } else {
+          briefing =
+            "Warning: server agent briefing unavailable; using bundled fallback.\n\n" +
+            readBriefing();
+        }
       }
     }
     return {
       messages: [
         {
           role: "user",
-          content: { type: "text", text: stub ?? readBriefing() },
+          content: { type: "text", text: stub ?? briefing ?? readBriefing() },
         },
       ],
     };

package/bin/llama.mjs

@@ -1,5 +1,6 @@
 #!/usr/bin/env node
 
+import { createRequire } from "module";
 import readline from "readline";
 import {
   DEFAULT_BASE_URL,
@@ -9,6 +10,7 @@ import {
   TOKEN_FILE,
   getAuthHeaders,
   getBaseUrl,
+  getLastAgentEvent,
   getToken,
   print,
   readBriefing,
@@ -33,6 +35,9 @@ import { LLAMA_CLI_CLIENT_ID, pkceLoopbackFlow, revokeToken as revokeOAuthToken
 import { deleteBundle, detectBackend, readBundle, writeBundle } from "../lib/oauth-storage.mjs";
 import { maybeNudgeUpdate, getUpdateNudge } from "../lib/version-check.mjs";
 
+const requireFromHere = createRequire(import.meta.url);
+const { version: PKG_VERSION } = requireFromHere("../package.json");
+
 function parseFlags(args, knownFlags = null) {
   const flags = {};
   const positional = [];
@@ -71,6 +76,34 @@ function parseFlags(args, knownFlags = null) {
   return { flags, positional };
 }
 
+function agentOnboardNoAuthMessage() {
+  return `Llama Ventures team onboarding requires credentials.
+
+Team member?
+  - Run \`gcloud auth login\` with your @llamaventures.vc account, OR
+  - Mint a token at https://command.llamaventures.vc/settings/tokens
+    then \`llama token set <llc_...>\`.
+  Re-run \`llama agent-onboard\` after — the workflow contract will print.
+
+Founder or external visitor (no Llama account)?
+  Run \`llama pitch start --name "Your Name" --email "you@company.com"\`
+  to chat with our intake agent — no token required.`;
+}
+
+function agentOnboardRejectedMessage() {
+  return `Llama Ventures team onboarding requires valid credentials.
+
+Server rejected the credentials we sent. Re-mint at
+https://command.llamaventures.vc/settings/tokens, run
+\`llama token set <llc_...>\`, then re-run \`llama agent-onboard\`.`;
+}
+
+async function fetchServerAgentBriefing() {
+  const params = new URLSearchParams({ clientVersion: PKG_VERSION });
+  const result = await request("GET", `/api/agent/briefing?${params}`);
+  return result?.briefing || "";
+}
+
 function closestKnownFlag(input, candidates) {
   let best = null;
   let bestScore = Infinity;
@@ -127,6 +160,45 @@ function slugifyTitle(title) {
   return slug;
 }
 
+function parseExpectedIds(raw) {
+  const text = typeof raw === "string" ? raw : "";
+  const expected = { dealIds: [], wikiSlugs: [], raw: [] };
+  for (const item of text.split(",").map((s) => s.trim()).filter(Boolean)) {
+    const [kind, ...rest] = item.split(":");
+    const value = rest.join(":").trim();
+    if (kind === "deal" && value) expected.dealIds.push(value);
+    else if ((kind === "wiki" || kind === "slug") && value) expected.wikiSlugs.push(value);
+    else expected.raw.push(item);
+  }
+  return expected;
+}
+
+async function submitEvalFeedback(action, flags, queryText = "") {
+  const useLast = flags.last !== false;
+  const last = useLast ? getLastAgentEvent() : null;
+  const eventId =
+    flags.event && flags.event !== true
+      ? Number(flags.event)
+      : last?.lastEventId ?? null;
+  if ((action === "good" || action === "bad") && !eventId && !queryText) {
+    throw new Error(`Usage: llama eval ${action} [--last] [--reason "..."]`);
+  }
+  const body = {
+    action,
+    eventId: Number.isFinite(eventId) ? eventId : undefined,
+    query: queryText || undefined,
+    surface: flags.surface && flags.surface !== true ? String(flags.surface) : last?.lastSurface ?? undefined,
+    expected:
+      flags.expect && flags.expect !== true
+        ? parseExpectedIds(String(flags.expect))
+        : {},
+    reason: flags.reason && flags.reason !== true ? String(flags.reason) : undefined,
+    privacyLevel:
+      flags.privacy && flags.privacy !== true ? String(flags.privacy) : "internal",
+  };
+  return request("POST", "/api/agent/eval-feedback", body);
+}
+
 // Client-side fuzzy match — used as a fallback when the server hasn't yet
 // shipped the search/filter API (Fix B, 2026-04-25). Once the server
 // returns the `{deals,total,limit,offset}` envelope, this path is never
@@ -282,6 +354,8 @@ Agent onboarding (run once on first install):
   llama skills search "pipeline update" # discover relevant runtime skills
   llama skills show llama-pipeline      # read a skill from Command
   llama explain <url-or-object>         # explain Command URL/object status + lifecycle
+  llama eval good|bad --last            # mark the latest CLI/MCP result for eval
+  llama eval add "<query>" --expect wiki:<slug>|deal:<uuid>
 
 External pitch — talk to Llama Ventures' intake agent (no token required):
   llama pitch start --name "Jane Doe" --email "jane@acme.ai"
@@ -525,6 +599,7 @@ Common:
   llama agent bootstrap             live Llama OS skill manifest from Command
   llama skills search "<query>"     discover which skill to read
   llama explain <url-or-object>     explain Command URLs, 404s, deleted objects
+  llama eval bad --last              mark latest CLI/MCP result as an eval candidate
 
 Command groups — run \`llama help <group>\` for that group's commands:
   deal        create · show · feed · update · enrich · search · collaborators · links · delete
@@ -532,6 +607,7 @@ Command groups — run \`llama help <group>\` for that group's commands:
   facts       deal facts + skill corrections (the sourced, trust-rated layer)
   timeline    timeline · posts · mentions
   wiki        cross-deal knowledge entries (markdown or HTML)
+  eval        mark real CLI/MCP searches good/bad or add a golden-query candidate
   memo        long-form HTML investment memo
   html        deal-specific HTML artifacts (/deals/<id>/browse/<slug>)
   pitch       external founder intake (no token needed)
@@ -865,18 +941,15 @@ async function runPitchRepl() {
 async function main() {
   const [area, action, ...rest] = process.argv.slice(2);
   if (area === "--version" || area === "-v" || area === "version") {
-    const { createRequire } = await import("module");
-    const requireFromHere = createRequire(import.meta.url);
-    const { version } = requireFromHere("../package.json");
     // `llama version --check` — explicitly check npm for a newer release and
     // print the upgrade line (or "up to date"). Lets an agent surface the
     // nudge on demand, separate from the throttled, TTY-gated auto-nudge.
     if (action === "--check" || action === "check") {
       const nudge = await getUpdateNudge();
-      console.log(nudge || `llama CLI ${version} — up to date`);
+      console.log(nudge || `llama CLI ${PKG_VERSION} — up to date`);
       return;
     }
-    console.log(version);
+    console.log(PKG_VERSION);
     return;
   }
   if (!area || area === "help" || area === "--help" || area === "-h") {
@@ -897,12 +970,13 @@ async function main() {
     return;
   }
 
-  // `llama agent-onboard` — print the bundled AGENT_BRIEFING.md so an AI
-  // agent reads it once and internalises the Llama Ventures workflow
-  // contract. Same content the `agent_briefing` MCP prompt returns.
+  // `llama agent-onboard` — fetch the server-owned Agent Runtime Contract
+  // so an AI agent reads the current Llama Ventures workflow contract. The
+  // bundled AGENT_BRIEFING.md is now only a fallback when the server route
+  // is unavailable during rollout.
   // Also: `llama agent onboard` (two-word form) for symmetry.
   //
-  // Gated behind /api/me — without valid credentials we print a short
+  // Gated behind Command auth — without valid credentials we print a short
   // bootstrap stub instead. Stops unauthenticated callers from harvesting
   // internal command surface / workflow conventions just by running the
   // public CLI.
@@ -912,39 +986,24 @@ async function main() {
   ) {
     const headers = await getAuthHeaders();
     if (Object.keys(headers).length === 0) {
-      console.log(
-`Llama Ventures team onboarding requires credentials.
-
-Team member?
-  - Run \`gcloud auth login\` with your @llamaventures.vc account, OR
-  - Mint a token at https://command.llamaventures.vc/settings/tokens
-    then \`llama token set <llc_...>\`.
-  Re-run \`llama agent-onboard\` after — the workflow contract will print.
-
-Founder or external visitor (no Llama account)?
-  Run \`llama pitch start --name "Your Name" --email "you@company.com"\`
-  to chat with our intake agent — no token required.`
-      );
+      console.log(agentOnboardNoAuthMessage());
       return;
     }
     try {
-      await request("GET", "/api/me");
+      const briefing = await fetchServerAgentBriefing();
+      process.stdout.write(briefing || readBriefing());
     } catch (e) {
       const msg = e?.message || "";
       if (msg.includes("Error[UNAUTHORIZED]") || msg.includes("Error[NO_AUTH]")) {
-        console.log(
-`Llama Ventures team onboarding requires valid credentials.
-
-Server rejected the credentials we sent. Re-mint at
-https://command.llamaventures.vc/settings/tokens, run
-\`llama token set <llc_...>\`, then re-run \`llama agent-onboard\`.`
-        );
+        console.log(agentOnboardRejectedMessage());
         process.exitCode = 1;
         return;
       }
-      throw e;
+      process.stderr.write(
+        `warning: server agent briefing unavailable (${msg}); using bundled fallback.\n`,
+      );
+      process.stdout.write(readBriefing());
     }
-    process.stdout.write(readBriefing());
     return;
   }
 
@@ -954,6 +1013,7 @@ https://command.llamaventures.vc/settings/tokens, run
   if (area === "agent" && action === "bootstrap") {
     const { flags } = parseFlags(rest, ["json", "limit"]);
     const params = new URLSearchParams();
+    params.set("clientVersion", PKG_VERSION);
     if (flags.limit && flags.limit !== true) params.set("limit", String(flags.limit));
     const manifest = await request("GET", `/api/agent/manifest${params.toString() ? `?${params}` : ""}`);
     if (flags.json) {
@@ -1037,6 +1097,43 @@ https://command.llamaventures.vc/settings/tokens, run
     return;
   }
 
+  if (area === "eval") {
+    const sub = action;
+    if (sub === "good" || sub === "bad") {
+      const { flags, positional } = parseFlags(rest, [
+        "last",
+        "event",
+        "reason",
+        "expect",
+        "surface",
+        "privacy",
+      ]);
+      const q = positional.join(" ").trim();
+      print(await submitEvalFeedback(sub, flags, q));
+      return;
+    }
+    if (sub === "add") {
+      const { flags, positional } = parseFlags(rest, [
+        "event",
+        "expect",
+        "reason",
+        "surface",
+        "privacy",
+      ]);
+      const q = positional.join(" ").trim();
+      if (!q && !flags.event) {
+        throw new Error(
+          `Usage: llama eval add "<query>" --surface deal|wiki --expect wiki:<slug>|deal:<uuid>`,
+        );
+      }
+      print(await submitEvalFeedback("add", flags, q));
+      return;
+    }
+    throw new Error(
+      "Usage: llama eval good|bad [--last] [--reason ...] OR llama eval add \"<query>\" --expect wiki:<slug>|deal:<uuid>",
+    );
+  }
+
   // `llama pitch ...` — external founder-pitch family. No Llama token
   // required; bootstraps a session against /api/external/* via PoW + cookie.
   // See lib/external.mjs and AGENT_BRIEFING.md for the full surface.

package/lib/client.mjs

@@ -13,6 +13,7 @@ import path from "path";
 import { fileURLToPath } from "url";
 import { execFile as _execFile } from "child_process";
 import { promisify } from "util";
+import { randomUUID } from "crypto";
 
 const execFile = promisify(_execFile);
 
@@ -43,6 +44,19 @@ export function readBriefing() {
   }
 }
 
+let packageVersionCache = null;
+
+export function getPackageVersion() {
+  if (packageVersionCache) return packageVersionCache;
+  try {
+    const pkg = JSON.parse(fs.readFileSync(path.join(PACKAGE_ROOT, "package.json"), "utf8"));
+    packageVersionCache = String(pkg.version || "unknown");
+  } catch {
+    packageVersionCache = "unknown";
+  }
+  return packageVersionCache;
+}
+
 // Canonical entrypoint. `llama-command.onrender.com` also serves the API
 // but its NextAuth callback URL doesn't match, so browser login (needed
 // to mint a token at /settings/tokens) fails there with a server-config
@@ -54,6 +68,7 @@ export const DEFAULT_BASE_URL = "https://command.llamaventures.vc";
 // agent-discovery convention.
 export const TOKEN_DIR = path.join(os.homedir(), ".llama");
 export const TOKEN_FILE = path.join(TOKEN_DIR, "token");
+export const AGENT_SESSION_FILE = path.join(TOKEN_DIR, "agent-session.json");
 
 // Legacy location used by CLI v0.1. Read for back-compat (silent migrate
 // to canonical on first use); never written for the token, but still the
@@ -123,6 +138,269 @@ export function getToken() {
   return "";
 }
 
+let runtimeClient = "cli";
+let runtimeAgentClient = null;
+
+export function setClientRuntime(opts = {}) {
+  if (opts.client) runtimeClient = String(opts.client);
+  if (opts.agentClient) runtimeAgentClient = String(opts.agentClient);
+}
+
+function detectAgentClient() {
+  if (runtimeAgentClient) return runtimeAgentClient;
+  if (process.env.LLAMA_AGENT_CLIENT) return process.env.LLAMA_AGENT_CLIENT;
+  if (process.env.CODEX_SANDBOX || process.env.CODEX_CLI || process.env.OPENAI_CODEX) return "codex";
+  if (process.env.CLAUDECODE || process.env.CLAUDE_CODE || process.env.CLAUDE_CODE_ENTRYPOINT) {
+    return "claude-code";
+  }
+  if (process.env.CURSOR_AGENT || process.env.CURSOR_TRACE_ID) return "cursor";
+  return "unknown";
+}
+
+export function readAgentSession() {
+  try {
+    return JSON.parse(fs.readFileSync(AGENT_SESSION_FILE, "utf8"));
+  } catch {
+    return {};
+  }
+}
+
+function writeAgentSession(session) {
+  try {
+    fs.mkdirSync(TOKEN_DIR, { recursive: true, mode: 0o700 });
+    fs.writeFileSync(AGENT_SESSION_FILE, `${JSON.stringify(session, null, 2)}\n`, { mode: 0o600 });
+    fs.chmodSync(AGENT_SESSION_FILE, 0o600);
+  } catch {
+    // Telemetry state is best-effort. Never break the actual CLI command.
+  }
+}
+
+function currentAgentSessionId() {
+  const session = readAgentSession();
+  if (session.sessionId) return session.sessionId;
+  const created = {
+    sessionId: randomUUID(),
+    createdAt: new Date().toISOString(),
+  };
+  writeAgentSession(created);
+  return created.sessionId;
+}
+
+export function getLastAgentEvent() {
+  const session = readAgentSession();
+  return session.lastEventId ? session : null;
+}
+
+function rememberAgentEvent(event) {
+  if (!event?.eventId) return;
+  const session = {
+    ...readAgentSession(),
+    sessionId: event.sessionId || currentAgentSessionId(),
+    lastEventId: event.eventId,
+    lastCandidateId: event.candidateId ?? null,
+    lastCommand: event.command ?? null,
+    lastQuery: event.query ?? null,
+    lastSurface: event.surface ?? null,
+    lastRecordedAt: new Date().toISOString(),
+  };
+  writeAgentSession(session);
+}
+
+function agentClientHeaders(command) {
+  return {
+    "X-Llama-Client": runtimeClient,
+    "X-Llama-Client-Version": getPackageVersion(),
+    "X-Llama-Agent-Client": detectAgentClient(),
+    "X-Llama-Agent-Session": currentAgentSessionId(),
+    "X-Llama-Command": command || "unknown",
+  };
+}
+
+const SECRET_KEY_RE = /(token|secret|password|authorization|cookie|api[_-]?key|keychain|jwt)/i;
+
+function truncateText(text, max = 2000) {
+  return text.length > max ? `${text.slice(0, max)}...[truncated]` : text;
+}
+
+function sanitizeTelemetryValue(value, depth = 0) {
+  if (depth > 4) return "[max-depth]";
+  if (value === null || value === undefined) return value;
+  if (typeof value === "string") return truncateText(value);
+  if (typeof value === "number" || typeof value === "boolean") return value;
+  if (Array.isArray(value)) {
+    return value.slice(0, 20).map((item) => sanitizeTelemetryValue(item, depth + 1));
+  }
+  if (typeof value === "object") {
+    const out = {};
+    for (const [key, val] of Object.entries(value).slice(0, 40)) {
+      out[key] = SECRET_KEY_RE.test(key) ? "[redacted]" : sanitizeTelemetryValue(val, depth + 1);
+    }
+    return out;
+  }
+  return String(value);
+}
+
+function parseEndpoint(endpoint) {
+  try {
+    return new URL(endpoint, "https://command.llamaventures.vc");
+  } catch {
+    return null;
+  }
+}
+
+function endpointArgs(endpoint, body) {
+  const args = {};
+  const url = parseEndpoint(endpoint);
+  if (url) {
+    for (const [key, value] of url.searchParams.entries()) args[key] = value;
+  }
+  if (body && typeof body === "object" && !Array.isArray(body)) {
+    Object.assign(args, body);
+  }
+  return sanitizeTelemetryValue(args);
+}
+
+function inferCommand(method, endpoint) {
+  const url = parseEndpoint(endpoint);
+  const pathname = url?.pathname || endpoint.split("?")[0] || "";
+  const verb = String(method || "GET").toUpperCase();
+  if (pathname === "/api/agent/client-events") return "telemetry.record";
+  if (pathname === "/api/agent/eval-feedback") return "eval.feedback";
+  if (pathname === "/api/wiki/search") return "wiki.search";
+  if (pathname === "/api/wiki/save") return "wiki.save";
+  if (/^\/api\/wiki\/[^/]+$/.test(pathname)) return verb === "GET" ? "wiki.read" : "wiki.write";
+  if (pathname === "/api/deals") return verb === "GET" ? "deal.search" : "deal.write";
+  if (pathname === "/api/deals/create") return "deal.create";
+  if (pathname === "/api/deals/update") return "deal.update";
+  if (/^\/api\/deals\/[^/]+\/threads\/[^/]+$/.test(pathname)) return "deal.agent.run";
+  if (/^\/api\/deals\/[^/]+\/threads$/.test(pathname)) return "deal.thread.create";
+  if (/^\/api\/deals\/[^/]+\/facts/.test(pathname)) return verb === "GET" ? "deal.fact.list" : "deal.fact.write";
+  if (/^\/api\/deals\/[^/]+\/posts$/.test(pathname)) return "deal.post";
+  if (/^\/api\/deals\/[^/]+\/blocks/.test(pathname)) return verb === "GET" ? "brief.blocks" : "brief.write";
+  if (/^\/api\/deals\/[^/]+$/.test(pathname)) return verb === "GET" ? "deal.show" : "deal.write";
+  if (pathname === "/api/me") return "auth.status";
+  if (pathname.startsWith("/api/agent/skills")) return "skills.read";
+  if (pathname === "/api/agent/manifest") return "agent.bootstrap";
+  if (pathname === "/api/agent/briefing") return "agent.briefing";
+  return `${verb.toLowerCase()} ${pathname || endpoint}`;
+}
+
+function queryForCommand(command, args) {
+  if (!command.endsWith(".search")) return null;
+  return args?.q || args?.search || args?.query || null;
+}
+
+function summarizeResultIds(data) {
+  const result = {};
+  const topDeals = [];
+  const topWiki = [];
+
+  const collectDeal = (deal) => {
+    const id = deal?.uuid || deal?.id || deal?.dealId || deal?.deal_uuid;
+    if (!id) return;
+    topDeals.push({ id, name: deal.companyName || deal.company_name || deal.name || null });
+  };
+  const collectWiki = (item) => {
+    if (!item?.slug) return;
+    topWiki.push({ slug: item.slug, title: item.title || null });
+  };
+
+  if (Array.isArray(data)) {
+    for (const item of data.slice(0, 20)) {
+      collectDeal(item);
+      collectWiki(item);
+    }
+    result.resultCount = data.length;
+  } else if (data && typeof data === "object") {
+    const deals = Array.isArray(data.deals) ? data.deals : [];
+    const articles = Array.isArray(data.articles) ? data.articles : [];
+    const results = Array.isArray(data.results) ? data.results : [];
+    for (const deal of deals.slice(0, 20)) collectDeal(deal);
+    for (const item of [...articles, ...results].slice(0, 20)) collectWiki(item);
+    if (typeof data.total === "number") result.total = data.total;
+    if (deals.length) result.resultCount = deals.length;
+    if (articles.length || results.length) result.resultCount = articles.length + results.length;
+  }
+
+  if (topDeals.length) result.deals = topDeals;
+  if (topWiki.length) result.wiki = topWiki;
+  return result;
+}
+
+function summarizeResult(data) {
+  if (data === null || data === undefined) return null;
+  if (Array.isArray(data)) return `${data.length} result(s)`;
+  if (typeof data === "object") {
+    if (Array.isArray(data.deals)) return `${data.deals.length} deal result(s); total=${data.total ?? "unknown"}`;
+    if (Array.isArray(data.results)) return `${data.results.length} result(s)`;
+    if (data.ok !== undefined) return `ok=${Boolean(data.ok)}`;
+  }
+  return truncateText(typeof data === "string" ? data : JSON.stringify(sanitizeTelemetryValue(data)), 2000);
+}
+
+function shouldSkipTelemetry(endpoint) {
+  if (process.env.LLAMA_TELEMETRY === "0") return true;
+  const pathname = parseEndpoint(endpoint)?.pathname || endpoint;
+  return pathname === "/api/agent/client-events" || pathname === "/api/agent/eval-feedback";
+}
+
+async function recordClientTelemetry({
+  authHeaders,
+  method,
+  endpoint,
+  body,
+  command,
+  status,
+  httpStatus,
+  latencyMs,
+  data,
+  errorMessage,
+}) {
+  if (shouldSkipTelemetry(endpoint)) return;
+  const args = endpointArgs(endpoint, body);
+  const sessionId = currentAgentSessionId();
+  const payload = {
+    client: runtimeClient,
+    clientVersion: getPackageVersion(),
+    agentClient: detectAgentClient(),
+    sessionId,
+    command,
+    method: String(method || "GET").toUpperCase(),
+    endpoint,
+    status,
+    httpStatus,
+    latencyMs,
+    args,
+    query: queryForCommand(command, args),
+    resultSummary: status === "success" ? summarizeResult(data) : null,
+    resultIds: status === "success" ? summarizeResultIds(data) : {},
+    errorMessage: errorMessage ? truncateText(String(errorMessage), 2000) : null,
+  };
+  try {
+    const res = await fetch(`${getBaseUrl()}/api/agent/client-events`, {
+      method: "POST",
+      headers: {
+        "Content-Type": "application/json",
+        ...agentClientHeaders("telemetry.record"),
+        ...authHeaders,
+      },
+      body: JSON.stringify(payload),
+    });
+    if (!res.ok) return;
+    const recorded = await res.json().catch(() => null);
+    rememberAgentEvent({
+      ...recorded,
+      sessionId,
+      command,
+      query: payload.query,
+      surface: command.startsWith("deal.") ? "deal" : command.startsWith("wiki.") ? "wiki" : null,
+    });
+  } catch {
+    // Best-effort by design. The actual llama command already succeeded or
+    // failed; telemetry must never alter that outcome.
+  }
+}
+
 // Try `gcloud auth print-identity-token`. Returns the JWT or null. Zero-config
 // win for any team member who has gcloud + their @llamaventures.vc account
 // already set up — the server's Bearer auth path verifies and auto-creates
@@ -227,10 +505,13 @@ export async function requestSse(method, endpoint, body, opts = {}) {
 async function requestWithRetry(method, endpoint, body, allowRetry) {
   const authHeaders = await getAuthHeaders();
   if (Object.keys(authHeaders).length === 0) throw noAuthError();
+  const command = inferCommand(method, endpoint);
+  const start = Date.now();
   const res = await fetch(`${getBaseUrl()}${endpoint}`, {
     method,
     headers: {
       "Content-Type": "application/json",
+      ...agentClientHeaders(command),
       ...authHeaders,
     },
     body: body === undefined ? undefined : JSON.stringify(body),
@@ -267,18 +548,45 @@ async function requestWithRetry(method, endpoint, body, allowRetry) {
   }
   if (!res.ok) {
     const message = typeof data === "object" && data?.error ? data.error : `HTTP ${res.status}`;
+    await recordClientTelemetry({
+      authHeaders,
+      method,
+      endpoint,
+      body,
+      command,
+      status: "error",
+      httpStatus: res.status,
+      latencyMs: Date.now() - start,
+      data: null,
+      errorMessage: message,
+    });
     throw new Error(message);
   }
+  await recordClientTelemetry({
+    authHeaders,
+    method,
+    endpoint,
+    body,
+    command,
+    status: "success",
+    httpStatus: res.status,
+    latencyMs: Date.now() - start,
+    data,
+    errorMessage: null,
+  });
   return data;
 }
 
 async function requestSseWithRetry(method, endpoint, body, opts, allowRetry) {
   const authHeaders = await getAuthHeaders();
   if (Object.keys(authHeaders).length === 0) throw noAuthError();
+  const command = inferCommand(method, endpoint);
+  const start = Date.now();
   const res = await fetch(`${getBaseUrl()}${endpoint}`, {
     method,
     headers: {
       "Content-Type": "application/json",
+      ...agentClientHeaders(command),
       ...authHeaders,
     },
     body: body === undefined ? undefined : JSON.stringify(body),
@@ -309,6 +617,18 @@ async function requestSseWithRetry(method, endpoint, body, opts, allowRetry) {
       data = text;
     }
     const message = typeof data === "object" && data?.error ? data.error : `HTTP ${res.status}`;
+    await recordClientTelemetry({
+      authHeaders,
+      method,
+      endpoint,
+      body,
+      command,
+      status: "error",
+      httpStatus: res.status,
+      latencyMs: Date.now() - start,
+      data: null,
+      errorMessage: message,
+    });
     throw new Error(message);
   }
 
@@ -343,6 +663,18 @@ async function requestSseWithRetry(method, endpoint, body, opts, allowRetry) {
     for (const frame of frames) handleFrame(frame);
   }
   if (buf.trim()) handleFrame(buf);
+  await recordClientTelemetry({
+    authHeaders,
+    method,
+    endpoint,
+    body,
+    command,
+    status: "success",
+    httpStatus: res.status,
+    latencyMs: Date.now() - start,
+    data: { ok: true, textLength: text.length, events: events.length },
+    errorMessage: null,
+  });
   return { text, events };
 }
 

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@llamaventures/cli",
-  "version": "1.15.1",
+  "version": "1.16.0",
   "description": "CLI + MCP server for the Llama Ventures investment workbench (command.llamaventures.vc).",
   "type": "module",
   "scripts": {

package/scripts/verify-agent-routing.mjs

@@ -22,6 +22,7 @@ assert.equal(
 );
 const calls = [];
 let threadSeq = 0;
+let eventSeq = 0;
 
 async function readJson(req) {
   let raw = "";
@@ -65,11 +66,47 @@ const server = createServer(async (req, res) => {
       path: url.pathname,
       query: Object.fromEntries(url.searchParams.entries()),
       body,
+      headers: {
+        client: req.headers["x-llama-client"] ?? null,
+        clientVersion: req.headers["x-llama-client-version"] ?? null,
+        agentClient: req.headers["x-llama-agent-client"] ?? null,
+        session: req.headers["x-llama-agent-session"] ?? null,
+        command: req.headers["x-llama-command"] ?? null,
+      },
     });
 
+    if (req.method === "POST" && url.pathname === "/api/agent/client-events") {
+      eventSeq += 1;
+      writeJson(res, {
+        ok: true,
+        eventId: eventSeq,
+        candidateId: body?.command?.endsWith(".search") ? eventSeq + 1000 : null,
+      });
+      return;
+    }
+
+    if (req.method === "POST" && url.pathname === "/api/agent/eval-feedback") {
+      writeJson(res, {
+        ok: true,
+        candidate: {
+          id: 42,
+          source_event_id: body?.eventId ?? null,
+          feedback: body?.action ?? null,
+        },
+      });
+      return;
+    }
+
     if (req.method === "GET" && url.pathname === "/api/agent/manifest") {
       writeJson(res, {
         ok: true,
+        contract: {
+          contract_version: "agent-contract.v1",
+          cli: {
+            client_version: url.searchParams.get("clientVersion"),
+            status: "ok",
+          },
+        },
         briefing: "runtime briefing: use skills_search, skills_read, and object_inspect",
         llama_os: {
           visible_skill_count: 49,
@@ -85,6 +122,21 @@ const server = createServer(async (req, res) => {
       return;
     }
 
+    if (req.method === "GET" && url.pathname === "/api/agent/briefing") {
+      writeJson(res, {
+        ok: true,
+        contract: {
+          contract_version: "agent-contract.v1",
+          cli: {
+            client_version: url.searchParams.get("clientVersion"),
+            status: "ok",
+          },
+        },
+        briefing: "server-owned briefing: check CLI, use Pipeline First, prefer CLI/MCP",
+      });
+      return;
+    }
+
     if (req.method === "GET" && url.pathname === "/api/agent/skills") {
       writeJson(res, {
         ok: true,
@@ -136,6 +188,16 @@ const server = createServer(async (req, res) => {
       return;
     }
 
+    if (req.method === "GET" && url.pathname === "/api/wiki/search") {
+      writeJson(res, [
+        {
+          slug: "llama-weekly-2026-06-16",
+          title: "Llama Weekly 2026-06-16",
+        },
+      ]);
+      return;
+    }
+
     if (req.method === "POST" && /^\/api\/deals\/[^/]+\/threads$/.test(url.pathname)) {
       threadSeq += 1;
       writeJson(res, { id: `thread-${threadSeq}` });
@@ -191,25 +253,34 @@ function resetCalls() {
   threadSeq = 0;
 }
 
+function businessCalls() {
+  return calls.filter((call) => call.path !== "/api/agent/client-events");
+}
+
+function telemetryCalls() {
+  return calls.filter((call) => call.path === "/api/agent/client-events");
+}
+
 function paths() {
-  return calls.map((call) => `${call.method} ${call.path}`);
+  return businessCalls().map((call) => `${call.method} ${call.path}`);
 }
 
 function assertNoEnrichCall() {
   assert.equal(
-    calls.some((call) => call.path.endsWith("/enrich")),
+    businessCalls().some((call) => call.path.endsWith("/enrich")),
     false,
     `expected no /enrich call, got ${paths().join(", ")}`,
   );
 }
 
 function assertThreadRun({ title, messageIncludes }) {
-  assert.equal(calls.length, 2, `expected thread create + SSE run, got ${paths().join(", ")}`);
-  assert.match(calls[0].path, /^\/api\/deals\/[^/]+\/threads$/);
-  assert.equal(calls[0].body?.title, title);
-  assert.match(calls[1].path, /^\/api\/deals\/[^/]+\/threads\/thread-1$/);
+  const relevant = businessCalls();
+  assert.equal(relevant.length, 2, `expected thread create + SSE run, got ${paths().join(", ")}`);
+  assert.match(relevant[0].path, /^\/api\/deals\/[^/]+\/threads$/);
+  assert.equal(relevant[0].body?.title, title);
+  assert.match(relevant[1].path, /^\/api\/deals\/[^/]+\/threads\/thread-1$/);
   for (const needle of messageIncludes) {
-    assert.match(calls[1].body?.message ?? "", new RegExp(escapeRegExp(needle)));
+    assert.match(relevant[1].body?.message ?? "", new RegExp(escapeRegExp(needle)));
   }
 }
 
@@ -313,18 +384,29 @@ const baseUrl = `http://${address.address}:${address.port}`;
 const homeDir = await mkdtemp(path.join(os.tmpdir(), "llama-cli-routing-"));
 
 try {
+  resetCalls();
+  const onboardRun = await runCli(["agent-onboard"], baseUrl, homeDir);
+  assert.match(onboardRun.stdout, /server-owned briefing/);
+  assert.deepEqual(paths(), ["GET /api/agent/briefing"]);
+  assert.ok(businessCalls()[0].query.clientVersion, "agent-onboard passes clientVersion");
+  assert.equal(telemetryCalls()[0].body?.command, "agent.briefing");
+  assert.equal(telemetryCalls()[0].body?.client, "cli");
+  assert.ok(telemetryCalls()[0].body?.sessionId, "telemetry includes an agent session id");
+  assert.equal(businessCalls()[0].headers.command, "agent.briefing");
+
   resetCalls();
   const bootstrapRun = await runCli(["agent", "bootstrap", "--limit", "3"], baseUrl, homeDir);
   assert.match(bootstrapRun.stdout, /runtime briefing/);
   assert.deepEqual(paths(), ["GET /api/agent/manifest"]);
-  assert.equal(calls[0].query.limit, "3");
+  assert.equal(businessCalls()[0].query.limit, "3");
+  assert.ok(businessCalls()[0].query.clientVersion, "agent bootstrap passes clientVersion");
 
   resetCalls();
   const skillSearchRun = await runCli(["skills", "search", "pipeline", "--limit", "5"], baseUrl, homeDir);
   assert.match(skillSearchRun.stdout, /llama-command/);
   assert.deepEqual(paths(), ["GET /api/agent/skills"]);
-  assert.equal(calls[0].query.q, "pipeline");
-  assert.equal(calls[0].query.limit, "5");
+  assert.equal(businessCalls()[0].query.q, "pipeline");
+  assert.equal(businessCalls()[0].query.limit, "5");
 
   resetCalls();
   const skillShowRun = await runCli(["skills", "show", "llama-command"], baseUrl, homeDir);
@@ -336,7 +418,34 @@ try {
   assert.match(explainRun.stdout, /Status: deleted/);
   assert.match(explainRun.stdout, /Deleted by Kevin Yu/);
   assert.deepEqual(paths(), ["GET /api/agent/explain"]);
-  assert.equal(calls[0].query.q, "https://command.llamaventures.vc/wiki/missing-page");
+  assert.equal(businessCalls()[0].query.q, "https://command.llamaventures.vc/wiki/missing-page");
+
+  resetCalls();
+  const wikiRun = await runCli(["wiki", "search", "llama weekly"], baseUrl, homeDir);
+  assert.match(wikiRun.stdout, /llama-weekly-2026-06-16/);
+  assert.deepEqual(paths(), ["GET /api/wiki/search"]);
+  assert.equal(telemetryCalls()[0].body?.command, "wiki.search");
+  assert.equal(telemetryCalls()[0].body?.query, "llama weekly");
+
+  resetCalls();
+  const evalRun = await runCli(
+    [
+      "eval",
+      "bad",
+      "--last",
+      "--reason",
+      "missed dev weekly",
+      "--expect",
+      "wiki:llamaos-weekly-2026-06-17",
+    ],
+    baseUrl,
+    homeDir,
+  );
+  assert.match(evalRun.stdout, /"feedback": "bad"/);
+  assert.deepEqual(paths(), ["POST /api/agent/eval-feedback"]);
+  assert.equal(businessCalls()[0].body?.action, "bad");
+  assert.equal(businessCalls()[0].body?.eventId, 6);
+  assert.equal(businessCalls()[0].body?.expected?.wikiSlugs?.[0], "llamaos-weekly-2026-06-17");
 
   resetCalls();
   const enrichRun = await runCli(
@@ -369,9 +478,9 @@ try {
     homeDir,
   );
   assert.deepEqual(paths(), ["POST /api/deals/deal-cli/enrich"]);
-  assert.equal(calls[0].body?.apply, true);
-  assert.equal(calls[0].body?.dryRun, false);
-  assert.equal(calls[0].body?.executor, "server_agent");
+  assert.equal(businessCalls()[0].body?.apply, true);
+  assert.equal(businessCalls()[0].body?.dryRun, false);
+  assert.equal(businessCalls()[0].body?.executor, "server_agent");
 
   resetCalls();
   const agentRun = await runCli(
@@ -414,14 +523,16 @@ try {
   const bootstrapPayload = JSON.parse(mcpBootstrap.content?.[0]?.text ?? "{}");
   assert.equal(bootstrapPayload.ok, true);
   assert.deepEqual(paths(), ["GET /api/agent/manifest"]);
-  assert.equal(calls[0].query.limit, "2");
+  assert.equal(businessCalls()[0].query.limit, "2");
+  assert.ok(businessCalls()[0].query.clientVersion, "mcp agent_bootstrap passes clientVersion");
+  assert.equal(telemetryCalls()[0].body?.client, "mcp");
 
   resetCalls();
   const mcpSkills = await callMcpTool("skills_search", { q: "command", limit: 4 }, baseUrl, homeDir);
   const skillsPayload = JSON.parse(mcpSkills.content?.[0]?.text ?? "{}");
   assert.equal(skillsPayload.skills?.[0]?.slug, "llama-command");
   assert.deepEqual(paths(), ["GET /api/agent/skills"]);
-  assert.equal(calls[0].query.q, "command");
+  assert.equal(businessCalls()[0].query.q, "command");
 
   resetCalls();
   const mcpSkillRead = await callMcpTool("skills_read", { slug: "llama-command" }, baseUrl, homeDir);