diff --git a/docs/public/robots.txt b/docs/public/robots.txt
index 340a0f4..597a269 100644
--- a/docs/public/robots.txt
+++ b/docs/public/robots.txt
@@ -1,3 +1,41 @@
+# neuron-js docs — agentic access policy (docs/public/robots.txt)
+# Served at https://sebasoft.github.io/neuron-js/robots.txt
+#
+# Decision (SebaSOFT, 2026-09): training crawlers are ALLOWED. All AI bots
+# listed below are explicitly welcomed so policy scanners read a clear
+# allow instead of inferring from a wildcard.
+
+# --- Citation-eligible search & answer bots ---
+User-agent: OAI-SearchBot
+Allow: /
+
+User-agent: PerplexityBot
+Allow: /
+
+User-agent: Claude-SearchBot
+Allow: /
+
+# --- User-triggered fetchers (agent browsing on a user's behalf) ---
+User-agent: ChatGPT-User
+Allow: /
+
+User-agent: Perplexity-User
+Allow: /
+
+User-agent: Claude-User
+Allow: /
+
+# --- Training crawlers (decision: ALLOWED) ---
+User-agent: GPTBot
+Allow: /
+
+User-agent: ClaudeBot
+Allow: /
+
+User-agent: Google-Extended
+Allow: /
+
+# --- Everything else (browsers, classic SEO crawlers, unlisted bots) ---
 User-agent: *
 Allow: /
 
diff --git a/tests/contracts/robots-granular.test.ts b/tests/contracts/robots-granular.test.ts
new file mode 100644
index 0000000..6e8a948
--- /dev/null
+++ b/tests/contracts/robots-granular.test.ts
@@ -0,0 +1,40 @@
+// NJS-SEO-6 follow-up: contract test for the granular robots.txt agentic
+// access policy. Pins the SebaSOFT decision (2026-09): training crawlers
+// are ALLOWED on the neuron-js documentation site.
+import { readFileSync } from "node:fs";
+import { describe, expect, it } from "vitest";
+
+const robots = readFileSync("docs/public/robots.txt", "utf8");
+
+describe("granular robots.txt agentic policy", () => {
+  it("explicitly allows citation-eligible AI bots", () => {
+    for (const bot of ["OAI-SearchBot", "PerplexityBot", "Claude-SearchBot"]) {
+      expect(robots).toContain(`User-agent: ${bot}`);
+    }
+  });
+
+  it("explicitly allows user-triggered fetchers", () => {
+    for (const bot of ["ChatGPT-User", "Perplexity-User", "Claude-User"]) {
+      expect(robots).toContain(`User-agent: ${bot}`);
+    }
+  });
+
+  it("pins the training-crawler decision: ALLOWED", () => {
+    for (const bot of ["GPTBot", "ClaudeBot", "Google-Extended"]) {
+      expect(robots).toContain(`User-agent: ${bot}`);
+      // every explicit rule in this file is an Allow — fail if someone flips policy
+      const section = robots.split(`User-agent: ${bot}`)[1]?.split("User-agent:")[0] ?? "";
+      expect(section).toContain("Allow: /");
+      expect(section).not.toContain("Disallow");
+    }
+    expect(robots).toContain("training crawlers are ALLOWED");
+  });
+
+  it("keeps the wildcard open and the sitemap for all agents", () => {
+    expect(robots).toContain("User-agent: *");
+    expect(robots).not.toContain("Disallow");
+    expect(robots).toContain(
+      "Sitemap: https://sebasoft.github.io/neuron-js/sitemap.xml",
+    );
+  });
+});