agentlinkops 0.6.9 → 0.6.10

package/.claude-plugin/plugin.json

@@ -1,6 +1,6 @@
 {
   "name": "agentlinkops",
-  "version": "0.6.9",
+  "version": "0.6.10",
   "description": "Connect an agent to AgentLinkOps and run backlink work with it: the connect skill (views, the three discovery verbs, evidence rules), local ledger and campaign skills, a portable CRM, the source-cited content pipeline, user-managed disavow lists, HTML reports and monitoring-event sync.",
   "author": {
     "name": "AgentLinkOps"

package/.codex-plugin/plugin.json

@@ -1,6 +1,6 @@
 {
   "name": "agentlinkops",
-  "version": "0.6.9",
+  "version": "0.6.10",
   "description": "Connect an agent to AgentLinkOps and run backlink work with it: the connect skill (views, the three discovery verbs, evidence rules), local ledger and campaign skills, a portable CRM, the source-cited content pipeline, user-managed disavow lists, HTML reports and monitoring-event sync.",
   "author": {
     "name": "AgentLinkOps"

package/CHANGELOG.md

@@ -3,6 +3,17 @@
 All notable changes to the `agentlinkops` package are recorded here. The version follows the
 plugin manifests in `.claude-plugin/` and `.codex-plugin/`.
 
+## 0.6.10 (2026-10-06)
+
+Inspect the installed version's help before using selectors; older published versions
+may install the full pack.
+
+- Default agent setup to the connection skill, its reference and detected-client connection setup. A connection-only upgrade retains previously installed optional payloads.
+- Select optional skills and versioned recipes with repeatable `--skill ID` and `--recipe ID`; choose the complete pack explicitly with `--all-skills`.
+- Refuse targeted skill removal while an installer-owned dependent recipe survives, including a preserved customer-edited body. Untrusted dependency metadata also refuses removal. Preview joint recipe/skill removal after reconciling edits, or preview a complete uninstall without selectors. Customer records and existing file/configuration custody rules remain in place.
+- Let customers choose their editorial process, tools and records. Publisher guidance and the bundled CRM are optional; selected recipe recovery no longer requires an unselected campaign recipe.
+- Make the supplied-link check visible before hosted examples. It fetches the source page without an account, repository, ledger, CRM or publisher recipe and retains observation uncertainty.
+
 ## 0.6.9 (2026-09-25)
 
 - Read ChatGPT citations in the current anonymous layout, where inline citations and source pills are buttons carrying their sources as JSON. 0.6.8 read only anchor links and recorded these answers as unknown. Code-copy, table-copy and entity controls are recognized as non-citations; any other unrecognized control still keeps the answer unknown.

package/README.md

@@ -12,28 +12,50 @@ separate service; this package is the local half and needs no account.
 
 ## Give your agent AgentLinkOps
 
-Four steps, the same ones the docs and the landing page show. Node.js 22.13 or later.
+Use Node.js 22.13 or later. Choose a supplied-link check or a hosted task, then inspect
+the installed version's help before selecting guidance.
 
-1. **Install the skill.** Detects the agent clients on this machine (Claude Code, Codex,
-   Cursor, Gemini CLI, Hermes), installs the skill pack at each one's skills path and writes
-   its MCP entry; stores no credential and prints what it changed.
+1. **Inspect setup and choose guidance.** CLI 0.6.10 installs the connection skill
+   and its reference by default, with MCP setup for detected clients. Select optional
+   guidance explicitly. Setup stores no credential and reports changes. CLI 0.6.9
+   and earlier can install the complete pack. Native-client acceptance remains
+   separate from package checks.
 
    ```sh
+   npx -y agentlinkops --help
    npx -y agentlinkops agent setup
    ```
 
    With the `skills` CLI instead: `npx -y skills add https://agentlinkops.com`, then add
    `https://app.agentlinkops.com/mcp` as a remote HTTP MCP server in your client.
-2. **Sign in when your agent asks.** The client opens the AgentLinkOps sign-in; you pick the
-   account, workspace and scopes. Nothing is pasted into a chat.
-3. **Verify.** Ask your agent to list your AgentLinkOps projects, or run `agentlinkops doctor`.
+   On a version whose help supports selection, use `agentlinkops agent setup --skill ID`
+   or `--recipe ID` for optional guidance; repeat either flag for several choices.
+   Choose `--all-skills` only for the complete pack. Customer instructions and records
+   remain the customer's choice.
+
+   Preview `agentlinkops agent remove --recipe ID` or `--skill ID` before adding
+   `--apply`. The installer refuses skill removal while a recipe it owns still
+   needs the skill. Keep the skill or select both that recipe and its skill after
+   reconciling customer edits. The installer preserves edited recipe bodies, which
+   can still block joint removal.
+   For an intended complete uninstall, preview
+   `agentlinkops agent remove --scope project` (or `user`), then add `--apply`.
+2. **Sign in for a chosen hosted task.** The client opens the AgentLinkOps sign-in;
+   you pick the account, workspace and scopes. Nothing is pasted into a chat.
+   A supplied-link check needs no account or hosted connection.
+3. **Verify hosted access when needed.** Ask your agent to list your AgentLinkOps
+   projects, or run `agentlinkops doctor`.
 4. **Check a placement.** Ask: "check whether [publisher page] links to [my page], show the
    dated observation and any uncertainty." No account yet? The same verifier runs locally:
 
    ```sh
-   npx -y agentlinkops check --source https://publisher.example/resources --target https://your-site.example/guide
+   npx -y agentlinkops check --source https://publisher.example/resources --target https://your-site.example/guide --scope exact --json
    ```
 
+   This fetches a public source page and returns an observation; no repository,
+   ledger, CRM or publisher recipe is required. Read the state and evidence limits,
+   including an unknown result. A returned result does not establish presence.
+
 The skill is a short door: the first time your agent uses AgentLinkOps in a session it reads
 the full agent reference, which `agentlinkops skill` prints and
 https://agentlinkops.com/SKILL.md serves. Hosted monitoring is an invitation-only pilot; the
@@ -50,11 +72,20 @@ agentlinkops --help
 Or run it once without installing:
 
 ```sh
-npx agentlinkops check
+npx -y agentlinkops check --source https://publisher.example/resources --target https://your-site.example/guide --scope exact --json
 ```
 
 ## Check one link with no account
 
+Check a supplied pair from any working directory, without creating a ledger:
+
+```sh
+agentlinkops check --source https://publisher.example/resources --target https://your-site.example/guide --scope exact --json
+```
+
+If you choose to retain expected links in a repository ledger, use the optional
+local workflow:
+
 ```sh
 cd your-repository
 agentlinkops init
@@ -67,9 +98,10 @@ agentlinkops status
 the source page and appends an observation to the mirror. `status` names every entry whose
 observations disagree with its intent and proposes the edit; applying it is a commit you make.
 
-Exit codes: `0` when every expected link is present, `1` only when an expected link is observed
-absent with complete evidence, `2` for a usage, configuration or ledger error. An unknown never
-fails the run. Pass `--fail-on-unknown` when your CI wants it to.
+For ledger checks, exit codes are `0` when every expected link is present, `1` only
+when an expected link is observed absent with complete evidence, and `2` for a usage,
+configuration or ledger error. An unknown never fails the run.
+Pass `--fail-on-unknown` when your CI wants it to.
 
 `agentlinkops doctor` verifies the ledger, the verifier self-test, cloud reachability and the
 token, and prints a fix line for anything that fails.

package/cli/admission.js

@@ -43,6 +43,7 @@ export async function admissionMain(
   argv,
   { client, out = console.log, err = console.error } = {},
 ) {
+  let json = false;
   try {
     const seen = new Set();
     for (const token of argv.filter((a) => a.startsWith("--"))) {
@@ -53,6 +54,7 @@ export async function admissionMain(
     }
     const a = parseArgs(argv, ["url"]),
       op = operations[a._[0]];
+    json = a.json === true;
     if (
       !op ||
       a._.length !== 1 ||
@@ -124,7 +126,13 @@ export async function admissionMain(
     out(JSON.stringify(result.data, null, 2));
     return 0;
   } catch (error) {
-    err(error?.code ? `${error.code}: ${error.message}` : error.message);
+    if (error.publicError) {
+      if (json) err(JSON.stringify({ error: error.publicError }));
+      else {
+        const next = [error.publicError.next, error.publicError.details?.next].find(value => typeof value === 'string' && value.trim());
+        err(`${error.code}: ${error.serverMessage ?? error.message}${next ? `\n${next}` : ''}`);
+      }
+    } else err(error?.code ? `${error.code}: ${error.message}` : error.message);
     return 2;
   }
 }

package/cli/agent-setup.js

@@ -25,11 +25,19 @@ import { NUDGE } from "./skill.js";
 // client. It is idempotent, never overwrites a file it did not write, and prints exactly what
 // changed. `agent status` reports what is installed without touching anything.
 const USAGE =
-  "agent setup [--client ID ...] [--scope project|user] [--origin URL] [--pack DIR] [--dry-run] [--json] | agent status [--json] | agent remove|recover [--scope project|user] [--apply] [--json]";
+  "agent setup [--client ID ...] [--scope project|user] [--origin URL] [--pack DIR] [--skill ID ...] [--recipe ID ...] [--all-skills] [--dry-run] [--json] | agent status [--json] | agent remove [--skill ID ...] [--recipe ID ...] [--scope project|user] [--apply] [--json] | agent recover [--scope project|user] [--apply] [--json]";
 export const DEFAULT_ORIGIN = "https://app.agentlinkops.com";
 export const SERVER_NAME = "agentlinkops";
 const RECEIPT = ".agentlinkops/agent-setup.json";
 const SHARED = Symbol("package siblings");
+const SELECTION = Symbol("selected payloads");
+const CORE_SKILL = "agentlinkops-connect";
+const selectionIds = (value, flag) => {
+  const ids = value === undefined ? [] : [].concat(value);
+  if (ids.some((id) => typeof id !== "string" || !/^[a-z][a-z0-9-]{0,79}$/u.test(id)))
+    throw new ConfigError(`--${flag} requires a named ID without traversal.`);
+  return [...new Set(ids)];
+};
 const ignored = (path) =>
   /(^|\/)(?:\._[^/]*|__pycache__|\.DS_Store)(?:\/|$)|\.(?:py[co]|sqlite(?:3)?(?:-[a-z]+)?|bak)$/u.test(
     path,
@@ -182,6 +190,136 @@ export async function readPack(dir) {
   return skills;
 }
 
+// Selection is additive: a core upgrade does not uninstall earlier opt-in payloads.
+// Explicit full-pack installation retains the legacy complete-pack upgrade behavior.
+export function selectPack(pack, { skills = [], recipes = [], all = false } = {}) {
+  if (all && (skills.length || recipes.length))
+    throw new ConfigError("Choose --all-skills or named --skill/--recipe selections.");
+  if (!Object.hasOwn(pack, CORE_SKILL) || !pack[CORE_SKILL]["SKILL.md"]?.bytes.length)
+    throw new ConfigError("The core connection skill is missing from the package.");
+  if (!pack[CORE_SKILL]["references/agentlinkops.md"]?.bytes.length)
+    throw new ConfigError("The core connection reference is missing from the package.");
+  if (all) return { pack, selection: { mode: "all", skills: Object.keys(pack), recipes: [] } };
+  const selected = new Set([CORE_SKILL, ...skills]);
+  const siblings = pack[SHARED] ?? {};
+  let catalog = null;
+  if (recipes.length) {
+    try { catalog = JSON.parse(siblings["references/recipes/catalog.json"]?.bytes.toString("utf8")); }
+    catch { throw new ConfigError("The package recipe catalog is unavailable or invalid."); }
+    if (catalog?.schemaVersion !== 1 || !Array.isArray(catalog.recipes))
+      throw new ConfigError("The package recipe catalog is unavailable or invalid.");
+    if (!siblings["references/recipes/command-notes.md"]?.bytes.length)
+      throw new ConfigError("The required recipe command notes are missing from the package.");
+  }
+  const chosenRecipes = recipes.map((id) => {
+    const matches = catalog.recipes.filter((row) => row?.id === id);
+    const row = matches[0];
+    if (matches.length !== 1 || row.path !== `${id}.md` ||
+        typeof row.version !== "string" || !/^\d+\.\d+\.\d+$/u.test(row.version) ||
+        typeof row.skill !== "string" || !Object.hasOwn(pack, row.skill) ||
+        !siblings[`references/recipes/${row.path}`])
+      throw new ConfigError(`Unknown or invalid recipe ${id}.`);
+    selected.add(row.skill);
+    return { id, version: row.version, skill: row.skill };
+  });
+  for (const id of selected)
+    if (!Object.hasOwn(pack, id)) throw new ConfigError(`Unknown skill ${id}.`);
+  const result = Object.fromEntries([...selected].map((id) => [id, pack[id]]));
+  const support = {};
+  // Optional growth skills retain their bundled support dependencies. Recipe bodies
+  // are separately selected; the catalog describes availability, not installation.
+  if ([...selected].some((id) => id !== CORE_SKILL))
+    for (const [name, data] of Object.entries(siblings))
+      if (!name.startsWith("references/recipes/")) support[name] = data;
+  if (recipes.length) {
+    for (const name of ["catalog.json", "command-notes.md", ...recipes.map((id) => `${id}.md`)]) {
+      const key = `references/recipes/${name}`;
+      if (siblings[key]) support[key] = siblings[key];
+    }
+  }
+  Object.defineProperty(result, SHARED, { value: support });
+  Object.defineProperty(result, SELECTION, { value: true });
+  return { pack: result, selection: { mode: "selected", skills: [...selected], recipes: chosenRecipes } };
+}
+
+// Targeted removal can operate with a historical receipt and no current package.
+// Shared support is retained because other installed skills may still depend on it.
+function selectedRemovalPath(path, skills, recipes, cwd, home) {
+  return AGENT_CLIENTS.some((client) => {
+    const bases = [client.skills.project && join(cwd, client.skills.project), expand(client.skills.user, home)].filter(Boolean);
+    return bases.some((base) => skills.some((id) => {
+      const rel = relative(join(base, id), path);
+      return rel && !rel.startsWith("..") && !isAbsolute(rel);
+    }) || recipes.some((id) => path === join(base, "..", "references", "recipes", `${id}.md`)));
+  });
+}
+
+// Derive dependencies from surviving owned payloads, not the most recent setup
+// selection: a later core-only upgrade deliberately retains earlier recipes.
+async function assertRemovalDependencies(receipt, operations, skills, cwd, home, scope) {
+  if (!skills.length) return;
+  const recovery = `Keep required skills, or preview removing each dependent recipe with its associated --skill ID after reconciling customer edits. For an intended complete uninstall, preview agent remove --scope ${scope}, then add --apply; no dependent files or MCP entries were changed.`;
+  const deleting = new Set(operations.filter((operation) => operation.bytes === null).map((operation) => operation.path));
+  const surviving = [];
+  for (const path of Object.keys(receipt.files)) {
+    if (deleting.has(path)) continue;
+    try {
+      await safePath(path);
+      if (!(await lstat(path)).isFile()) throw new Error("not a regular payload");
+      surviving.push(path);
+    } catch (error) {
+      if (error.code !== "ENOENT")
+        throw new ConfigError(`Cannot inspect surviving installer-owned payloads: an owned path is unsafe, unreadable or not a regular file. ${recovery}`);
+    }
+  }
+  const bases = new Set(AGENT_CLIENTS.flatMap((client) =>
+    [client.skills.project && join(cwd, client.skills.project), expand(client.skills.user, home)].filter(Boolean)));
+  const dependents = new Set();
+  for (const base of bases) {
+    if (skills.includes(CORE_SKILL)) {
+      for (const path of surviving) {
+        const rel = relative(base, path);
+        const id = rel.split(/[\\/]/u)[0];
+        if (rel && !rel.startsWith("..") && !isAbsolute(rel) && id !== CORE_SKILL)
+          dependents.add(`skill ${id} (requires ${CORE_SKILL})`);
+      }
+    }
+    const recipeRoot = join(base, "..", "references", "recipes");
+    const recipes = surviving.flatMap((path) => {
+      const match = /^([a-z][a-z0-9-]{0,79})\.md$/u.exec(relative(recipeRoot, path));
+      return match && match[1] !== "command-notes" ? [match[1]] : [];
+    });
+    if (!recipes.length) continue;
+    if (skills.includes(CORE_SKILL)) {
+      for (const id of recipes) dependents.add(`recipe ${id} (requires ${CORE_SKILL})`);
+      continue;
+    }
+    const catalogPath = join(recipeRoot, "catalog.json");
+    let catalog;
+    try {
+      if (!Object.hasOwn(receipt.files, catalogPath)) throw new Error("unowned catalog");
+      const catalogBytes = await bytesAt(catalogPath);
+      if (!catalogBytes || sha256(catalogBytes) !== receipt.files[catalogPath]) throw new Error("catalog custody changed");
+      catalog = JSON.parse(catalogBytes);
+      if (catalog?.schemaVersion !== 1 || !Array.isArray(catalog.recipes)) throw new Error("invalid catalog");
+    } catch {
+      throw new ConfigError(`Cannot determine dependencies of surviving recipes ${recipes.join(", ")}: the owned recipe catalog is missing, edited or invalid. ${recovery}`);
+    }
+    for (const id of recipes) {
+      const matches = catalog.recipes.filter((row) => row?.id === id);
+      const row = matches[0];
+      if (matches.length !== 1 || row.path !== `${id}.md` || typeof row.skill !== "string" ||
+          !/^[a-z][a-z0-9-]{0,79}$/u.test(row.skill))
+        throw new ConfigError(`Cannot determine the required skill for surviving recipe ${id}: its owned catalog entry is missing or ambiguous. ${recovery}`);
+      if (skills.includes(row.skill)) dependents.add(`recipe ${id} (requires skill ${row.skill}; select --recipe ${id} with --skill ${row.skill})`);
+    }
+  }
+  if (skills.includes(CORE_SKILL) && Object.keys(receipt.mcp ?? {}).length)
+    dependents.add("retained MCP connection ownership");
+  if (dependents.size)
+    throw new ConfigError(`Removal would leave surviving dependents or MCP connection ownership: ${[...dependents].sort().join("; ")}. ${recovery}`);
+}
+
 const expand = (path, home) =>
   path.startsWith("~/") ? join(home, path.slice(2)) : path;
 const exists = async (path) => !!(await lstat(path).catch(() => null));
@@ -445,7 +583,7 @@ export async function planSkills(client, pack, { cwd, home, scope, receipt }) {
     unchanged: [],
     conflicts: [],
     desired: [],
-    roots: [
+    roots: pack[SELECTION] ? Object.keys(pack).map((name) => join(dir, name)) : [
       dir,
       ...["scripts", "references", "templates"].map((name) =>
         join(dir, "..", name),
@@ -764,7 +902,7 @@ export async function agentMain(
     afterOperation,
   } = {},
 ) {
-  const args = parseArgs(argv, ["client"]);
+  const args = parseArgs(argv, ["client", "skill", "recipe"]);
   const [command, sub] = args._;
   if (
     command !== "agent" ||
@@ -774,24 +912,34 @@ export async function agentMain(
     throw new ConfigError(USAGE);
   const allowed =
     sub === "setup"
-      ? ["_", "client", "scope", "origin", "pack", "dry-run", "json"]
-      : ["remove", "recover"].includes(sub)
-        ? ["_", "scope", "apply", "json"]
+      ? ["_", "client", "scope", "origin", "pack", "skill", "recipe", "all-skills", "dry-run", "json"]
+      : sub === "remove"
+        ? ["_", "scope", "apply", "skill", "recipe", "json"]
+        : sub === "recover" ? ["_", "scope", "apply", "json"]
         : ["_", "json"];
   if (Object.keys(args).some((k) => !allowed.includes(k)))
     throw new ConfigError(USAGE);
+  const skillIds = selectionIds(args.skill, "skill");
+  const recipeIds = selectionIds(args.recipe, "recipe");
+  if (args["all-skills"] !== undefined && args["all-skills"] !== true)
+    throw new ConfigError("--all-skills is a boolean flag.");
   const json = args.json === true || !isTTY;
   const scope = args.scope ?? "project";
   if (!["project", "user"].includes(scope))
     throw new ConfigError("--scope is project or user");
-  const origin = (
-    args.origin ??
-    env.AGENTLINKOPS_API_URL ??
-    env.LINKTRAIL_API_URL ??
-    DEFAULT_ORIGIN
-  ).replace(/\/$/u, "");
-  if (!/^https:\/\/[^/?#]+$/u.test(origin))
-    throw new ConfigError("--origin must be an https origin without a path.");
+  const suppliedOrigin = args.origin ?? env.AGENTLINKOPS_API_URL ?? env.LINKTRAIL_API_URL ?? DEFAULT_ORIGIN;
+  let origin;
+  try {
+    if (typeof suppliedOrigin !== "string" || /[\u0000-\u0020\u007f"'\\@]/u.test(suppliedOrigin))
+      throw new Error("invalid origin");
+    const parsed = new URL(suppliedOrigin);
+    if (parsed.protocol !== "https:" || parsed.username || parsed.password || parsed.search || parsed.hash || parsed.pathname !== "/" ||
+        (suppliedOrigin !== parsed.origin && suppliedOrigin !== `${parsed.origin}/`))
+      throw new Error("invalid origin");
+    origin = parsed.origin;
+  } catch {
+    throw new ConfigError("--origin must be a canonical https origin without credentials, controls or a path.");
+  }
   const only = args.client ? [].concat(args.client) : null;
   for (const id of only ?? [])
     if (!clientById(id))
@@ -818,9 +966,18 @@ export async function agentMain(
   }
   const receipt = await readReceipt(paths.receipt, cwd, home);
   if (sub === "remove") {
-    const operations = [],
-      preserved = [];
+    const targeted = skillIds.length > 0 || recipeIds.length > 0;
+    const operations = [], preserved = [], retained = [];
+    for (const id of [...skillIds, ...recipeIds]) {
+      const found = Object.keys(receipt?.files ?? {}).some((path) => selectedRemovalPath(path,
+        skillIds.includes(id) ? [id] : [], recipeIds.includes(id) ? [id] : [], cwd, home));
+      if (!found) throw new ConfigError(`Selection ${id} is not owned by this installation.`);
+    }
     for (const [path, digest] of Object.entries(receipt?.files ?? {})) {
+      if (targeted && !selectedRemovalPath(path, skillIds, recipeIds, cwd, home)) {
+        retained.push(path);
+        continue;
+      }
       const bytes = await bytesAt(path);
       if (bytes === null) continue;
       if (sha256(bytes) !== digest) {
@@ -829,9 +986,10 @@ export async function agentMain(
       }
       operations.push({ path, bytes: null, expected: digest });
     }
+    if (targeted) await assertRemovalDependencies(receipt, operations, skillIds, cwd, home, scope);
     const remainingMcp = { ...(receipt?.mcp ?? {}) };
     const configurations = [];
-    for (const [path, record] of Object.entries(remainingMcp)) {
+    for (const [path, record] of Object.entries(targeted ? {} : remainingMcp)) {
       const planned = await planMcpRemoval(path, record);
       if (planned.preserved) {
         preserved.push(path);
@@ -845,7 +1003,7 @@ export async function agentMain(
     }
     const remaining = Object.fromEntries(
       Object.entries(receipt?.files ?? {}).filter(([p]) =>
-        preserved.includes(p),
+        preserved.includes(p) || retained.includes(p),
       ),
     );
     if (receipt) {
@@ -875,6 +1033,7 @@ export async function agentMain(
             .filter((o) => o.path !== paths.receipt && o.bytes === null)
             .map((o) => o.path),
           preserved,
+          retained,
           configurations,
         },
         null,
@@ -912,17 +1071,21 @@ export async function agentMain(
       ? {
           installedAt: receipt.installedAt,
           files: Object.keys(receipt.files).length,
+          ...(receipt.lastSetupSelection ? { lastSetupSelection: receipt.lastSetupSelection } : {}),
         }
       : null;
     out(json ? JSON.stringify(report, null, 2) : renderStatus(report));
     return 0;
   }
 
-  const pack = await readPack(await locatePack(args.pack ?? packDir));
+  const { pack, selection } = selectPack(await readPack(await locatePack(args.pack ?? packDir)), {
+    skills: skillIds, recipes: recipeIds, all: args["all-skills"] === true,
+  });
   const clients = await detectClients(home, only);
   const report = {
     origin,
     scope,
+    selection,
     dryRun: args["dry-run"] === true,
     clients: [],
     next: [],
@@ -999,6 +1162,8 @@ export async function agentMain(
       report.next.push(`${client.name}: ${c.target} ${c.note}`);
     report.clients.push(entry);
   }
+  report.retainedExisting = Object.keys(receipt?.files ?? {}).filter((path) => !desired.has(path) &&
+    ![...selectedRoots].some((root) => { const rel = relative(root, path); return rel && !rel.startsWith("..") && !isAbsolute(rel); }));
   report.pruned = [];
   report.preservedObsolete = [];
   let ownershipChanged = false;
@@ -1039,6 +1204,7 @@ export async function agentMain(
             origin,
             scope,
             clients: report.clients.map((c) => c.id),
+            lastSetupSelection: selection,
             files: ownedFiles,
             mcp: ownedMcp,
           },
@@ -1070,8 +1236,11 @@ export async function agentMain(
     }
   }
   report.next.push(
-    "Sign in through each client's MCP OAuth flow; no credential was stored by this command.",
+    "For a chosen hosted task, sign in through the client's MCP OAuth flow. Local supplied-link checks require no AgentLinkOps account; no credential was stored by this command.",
   );
+  if (report.retainedExisting.length)
+    report.next.push("Earlier payloads outside this selection were retained. Preview agent remove --skill ID or --recipe ID to remove them explicitly.");
+  report.next.push("Optional skills and recipes: agent setup --skill ID or --recipe ID; --all-skills installs the complete pack.");
   report.next.push(NUDGE);
   out(json ? JSON.stringify(report, null, 2) : renderSetup(report));
   return report.clients.some(