fix(imessage): recognize MiniMax mm: reasoning tags in reflection guard (completes #93767) (#93820)

Alix-007 · Peter Steinberger · web-flow · commit 076da567f434 · 2026-06-30T12:41:42.000-07:00
* fix(imessage): detect MiniMax mm: namespaced reasoning tags in reflection guard The iMessage reflection guard's THINKING_TAG_RE only matched bare `<think>/<thought>` tags, so MiniMax's `<mm:think>` namespaced reasoning tags (and Anthropic's `antml:` prefix) were not recognized as reflected assistant content. When such reasoning leaked back inbound, the guard treated it as a normal user message instead of dropping it, allowing recursive echo amplification. Accept the known reasoning-tag namespace prefixes `(?:antml:|mm:)?` on the think/thought alternatives, mirroring the shared reasoning-tag contract in src/shared/text/reasoning-tags.ts (PR #93767). This is the remaining channel-monitor complement of #93767 for the iMessage path; the optional prefix is a pure recognition enhancement with no change to the bare-tag, code-region exemption, or other reflection patterns. Verified by driving the real detectReflectedContent export: <mm:think>secret</mm:think>visible -> isReflection=true (was false) <think>...</think> -> isReflection=true <think>secret</think> -> isReflection=true (no regression) "thinking about minimax algorithms" -> isReflection=false (no false-positive) `<mm:think>x</mm:think>` in code -> isReflection=false (code exemption kept) * chore: retrigger CI for real behavior proof check * refactor(imessage): tighten reflection guard coverage --------- Co-authored-by: Peter Steinberger <steipete@golden-gate.local>
diff --git a/extensions/imessage/src/monitor/reflection-guard.test.ts b/extensions/imessage/src/monitor/reflection-guard.test.ts
@@ -37,6 +37,15 @@ describe("detectReflectedContent", () => {
     expect(result.matchedLabels).toContain("thinking-tag");
   });
 
+  it.each([
+    { provider: "Anthropic", tag: "antml:think" },
+    { provider: "MiniMax", tag: "mm:think" },
+  ])("detects $provider namespaced reasoning tags", ({ tag }) => {
+    const result = detectReflectedContent(`<${tag}>secret</${tag}>visible`);
+    expect(result.isReflection).toBe(true);
+    expect(result.matchedLabels).toContain("thinking-tag");
+  });
+
   it("detects <relevant_memories> tags", () => {
     const result = detectReflectedContent("<relevant_memories>data</relevant_memories>");
     expect(result.isReflection).toBe(true);
diff --git a/extensions/imessage/src/monitor/reflection-guard.ts b/extensions/imessage/src/monitor/reflection-guard.ts
@@ -4,7 +4,10 @@ import { findCodeRegions, isInsideCode } from "openclaw/plugin-sdk/text-chunking
 const INTERNAL_SEPARATOR_RE = /(?:#\+){2,}#?/;
 const ASSISTANT_ROLE_MARKER_RE = /\bassistant\s+to\s*=\s*\w+/i;
 // Require closing `>` to avoid false-positives on phrases like "<thought experiment>".
-const THINKING_TAG_RE = /<\s*\/?\s*(?:think(?:ing)?|thought|antthinking)\b[^<>]*>/i;
+// Keep known provider namespaces aligned with the public reasoning-tag contract
+// so reflected hidden reasoning cannot re-enter the agent loop.
+const THINKING_TAG_RE =
+  /<\s*\/?\s*(?:(?:antml:|mm:)?(?:think(?:ing)?|thought)|antthinking)\b[^<>]*>/i;
 const RELEVANT_MEMORIES_TAG_RE = /<\s*\/?\s*relevant[-_]memories\b[^<>]*>/i;
 // Require closing `>` to avoid false-positives on phrases like "<final answer>".
 const FINAL_TAG_RE = /<\s*\/?\s*final\b[^<>]*>/i;
