diff --git a/CHANGELOG.md b/CHANGELOG.md
index e558807..c9b2970 100644
--- a/CHANGELOG.md
+++ b/CHANGELOG.md
@@ -139,6 +139,8 @@ and adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0.html).
 
 #### MCP / indexing
 
+- `codegraph_explore` now makes clear that suggested call counts are advisory, so agents keep exploring when an answer is incomplete; thanks @rongbc. (#1504, #1570)
+
 - C++ functions following anonymous namespaces containing raw-string templates are now indexed correctly, even when template text resembles an unfinished macro call. (#1505)
 
 - Indexing now warns when parser errors leave a file with no symbols, including C++ raw strings with 16-character delimiters, so missing code is no longer silent. (#1522)
diff --git a/__tests__/explore-output-budget.test.ts b/__tests__/explore-output-budget.test.ts
index 9d9a0b3..efe302f 100644
--- a/__tests__/explore-output-budget.test.ts
+++ b/__tests__/explore-output-budget.test.ts
@@ -6,7 +6,7 @@
  * grep+Read. These tests pin the per-tier budget shape so future tuning
  * doesn't silently drift the small-project case back into bloat.
  */
-import { describe, it, expect, beforeAll, afterAll } from 'vitest';
+import { describe, it, expect, vi, beforeAll, afterAll } from 'vitest';
 import * as fs from 'fs';
 import * as path from 'path';
 import * as os from 'os';
@@ -198,7 +198,26 @@ describe('codegraph_explore output respects the adaptive budget', () => {
     const text = result.content?.[0]?.text ?? '';
     expect(text).not.toContain('### Additional relevant files');
     expect(text).not.toContain('Complete source code is included above');
-    expect(text).not.toContain('Explore budget:');
+    expect(text).not.toContain('advisory only, NOT a quota');
+  });
+
+  it('emits advisory-only exploration guidance on medium projects — never quota wording', async () => {
+    // Medium tier (500–4,999 files) turns the guidance note on. The synthetic
+    // project is tiny, so fake the stats to land in that tier — the note's
+    // WORDING is what this test pins. Regression guard: quota phrasing
+    // ("remaining calls" / "Synthesize once") must never come back — agents
+    // read it as a hard cap, stop exploring early, and fall back to grep+Read.
+    const spy = vi.spyOn(cg, 'getStats').mockReturnValue({ fileCount: 1000 } as ReturnType<CodeGraph['getStats']>);
+    try {
+      const result = await handler.execute('codegraph_explore', { query: 'Session method helper' });
+      const text = result.content?.[0]?.text ?? '';
+      expect(text).toContain('advisory only, NOT a quota');
+      expect(text).toContain('extra calls are never rejected or rate-limited');
+      expect(text).not.toContain('remaining calls');
+      expect(text).not.toContain('Synthesize once');
+    } finally {
+      spy.mockRestore();
+    }
   });
 
   it('still includes the Relationships section — it is the cheapest structural signal', async () => {
diff --git a/__tests__/mcp-tool-annotations.test.ts b/__tests__/mcp-tool-annotations.test.ts
index 2162e00..2b5920b 100644
--- a/__tests__/mcp-tool-annotations.test.ts
+++ b/__tests__/mcp-tool-annotations.test.ts
@@ -108,11 +108,12 @@ describe('Live tool surface keeps annotations with a project open (#1018)', () =
     expect(got.length).toBeGreaterThan(0);
     for (const tool of got) expectReadOnly(tool);
 
-    // explore's description is regenerated with a per-repo budget suffix via
-    // object spread; the annotation must survive that rewrite.
+    // explore's description is regenerated with a per-repo advisory-guidance
+    // suffix via object spread; the annotation must survive that rewrite.
     const explore = got.find((t) => t.name === 'codegraph_explore');
     expect(explore).toBeDefined();
-    expect(explore!.description).toMatch(/Budget: make at most/);
+    expect(explore!.description).toMatch(/advisory only, NOT a quota/);
+    expect(explore!.description).not.toMatch(/make at most/);
     expectReadOnly(explore!);
     expectExploreAlwaysLoad(got);
   });
diff --git a/scripts/agent-eval/probe-suite-envelope.mjs b/scripts/agent-eval/probe-suite-envelope.mjs
index dd04aad..aaf24d9 100644
--- a/scripts/agent-eval/probe-suite-envelope.mjs
+++ b/scripts/agent-eval/probe-suite-envelope.mjs
@@ -85,7 +85,7 @@ try {
       epilogueCut: text.includes('omitted for size'),
       sectionCut: text.includes('output truncated to budget'),
       notShown: text.includes('Not shown above'),
-      budgetNote: text.includes('**Explore budget:'),
+      budgetNote: text.includes('advisory only, NOT a quota'),
     });
   }
 } finally {
diff --git a/src/mcp/server-instructions.ts b/src/mcp/server-instructions.ts
index 2b2d4f8..f7e6e02 100644
--- a/src/mcp/server-instructions.ts
+++ b/src/mcp/server-instructions.ts
@@ -54,7 +54,7 @@ calls; a grep/read exploration is dozens.
 - **Almost any question — "how does X work", architecture, a bug, "what/where is X", or surveying an area** → \`codegraph_explore\` with a natural-language question or the relevant names. ONE capped call returns the verbatim source grouped by file; most often the ONLY call you need.
 - **"How does X reach/become Y? / the flow / the path from X to Y"** → \`codegraph_explore\`, naming the symbols that span the flow (e.g. \`mutateElement renderScene\`) — it surfaces the call path among them, riding dynamic-dispatch hops, and returns their source.
 - **Reading or editing a file/symbol you can name** → put its name or file path in the \`codegraph_explore\` query — it returns that current line-numbered source (safe to \`Edit\` from) with the call path and blast radius attached, so you don't Read it separately. For an overloaded name it returns every matching definition's body in one call.
-- **Need more?** Call \`codegraph_explore\` again with more specific names — treat the source it returns as already Read.
+- **Need more?** Call \`codegraph_explore\` again with more specific names — treat the source it returns as already Read. Suggested call counts are advisory only, NOT a quota; extra calls are never rejected or rate-limited.
 - Qualified symbol names accept dots, \`::\`, or slashes, including containers whose names contain dots (for example, \`AppWeb.Format.group\`).
 
 ## Anti-patterns
diff --git a/src/mcp/tools.ts b/src/mcp/tools.ts
index 8ef64a4..dd57d8a 100644
--- a/src/mcp/tools.ts
+++ b/src/mcp/tools.ts
@@ -207,7 +207,10 @@ export interface ExploreOutputBudget {
   includeAdditionalFiles: boolean;
   /** Include the "Complete source code is included above…" reminder. */
   includeCompletenessSignal: boolean;
-  /** Include the explore-budget reminder at the end. */
+  /**
+   * Include the advisory exploration-guidance note at the end. Purely
+   * advisory — the server NEVER rejects or rate-limits extra explore calls.
+   */
   includeBudgetNote: boolean;
 }
 
@@ -1686,7 +1689,7 @@ export class ToolHandler {
         if (tool.name === 'codegraph_explore') {
           return {
             ...tool,
-            description: `${tool.description} Budget: make at most ${budget} calls for this project (${stats.fileCount.toLocaleString()} files indexed).`,
+            description: `${tool.description} Exploration guidance — advisory only, NOT a quota: ~${budget} focused calls usually cover this project (${stats.fileCount.toLocaleString()} files indexed), and extra calls are never rejected or rate-limited.`,
           };
         }
         return tool;
@@ -5941,13 +5944,17 @@ export class ToolHandler {
         ? ['', `> Some file sections were trimmed for size. Elided symbols are named inside gap markers as \`name (file:line)\` and preferred in the file header — run another \`codegraph_explore\` (or \`codegraph_node\`) with those exact names for their source.`]
         : [];
 
-    // Explore budget note based on project size.
+    // Advisory exploration-guidance note based on project size. Deliberately
+    // phrased as guidance, NOT a quota: agents read "budget / remaining calls /
+    // Synthesize once" as a hard cap and stop exploring early, falling back to
+    // grep + Read (which costs more tokens). The server never rejects or
+    // rate-limits extra explore calls, and the note says so explicitly.
     let budgetBlock: string[] = [];
     if (budget.includeBudgetNote) {
       try {
         const stats = cg.getStats();
         const callBudget = getExploreBudget(stats.fileCount);
-        budgetBlock = ['', `> **Explore budget: ${callBudget} calls for this project (${stats.fileCount.toLocaleString()} files indexed).** Each call covers ~6 files; if your question spans more, spend your remaining calls on the uncovered area BEFORE falling back to Read — another explore is cheaper and more complete than reading those files. Synthesize once you've used ${callBudget}.`];
+        budgetBlock = ['', `> **Exploration guidance — advisory only, NOT a quota: this project (~${stats.fileCount.toLocaleString()} files indexed) is usually covered in ≈${callBudget} focused explore calls, and extra calls are never rejected or rate-limited.** If the response above does not fully cover your question, run another codegraph_explore on the uncovered symbols — it is cheaper and more complete than Read. Only stop exploring when the response actually covers the flow you asked about.`];
       } catch {
         // Stats unavailable — skip budget note
       }