Summary

• Problem: iMessage reflected replies can carry short attributed-text prefix bytes like \x93b\x02, which makes echo-cache text fallback miss the agent's own outbound text.
• Why it matters: when the reflected text misses echo matching, self-chat and direct-chat reply loops can re-enter the session and trigger duplicate or recursive replies.
• What changed: extensions/imessage/src/monitor/echo-cache.ts now strips the leading attributed-text marker pattern before newline/trim normalization, and targeted regression tests were added for cache matching plus the self-chat inbound decision path.
• What did NOT change (scope boundary): no routing changes, no TTL changes, no send-path changes, and no generic text sanitization outside the iMessage echo cache.

Change Type (select all)

• Bug fix
• Feature
• Refactor required for the fix
• Docs
• Security hardening
• Chore/infra

Scope (select all touched areas)

• Gateway / orchestration
• Skills / tool execution
• Auth / tokens
• Memory / storage
• Integrations
• API / contracts
• UI / DX
• CI/CD / infra

Linked Issue/PR

• Closes iMessage echo detection fails for messages with attributed text prefix bytes #64309
• Related iMessage: harden echo detection against control-char prefixes #59386
• This PR fixes a bug or regression

Root Cause (if applicable)

• Root cause: normalizeEchoTextKey() normalized newlines and trimming only, so reflected texts like "\x93b\x02Cached reply" did not match the clean outbound text stored in the echo cache.
• Missing detection / guardrail: there was no regression test covering attributed-text prefix bytes in the iMessage echo fallback path.
• Contributing context (if known): iMessage chat.db reflections can preserve short attributed-text marker bytes on some messages, especially richer/longer text.

Regression Test Plan (if applicable)

• Coverage level that should have caught this:
  • Unit test
  • Seam / integration test
  • End-to-end test
  • Existing coverage already sufficient
• Target test or file:
  • extensions/imessage/src/monitor/monitor-provider.echo-cache.test.ts
  • extensions/imessage/src/monitor/self-chat-dedupe.test.ts
• Scenario the test should lock in: a reflected iMessage body with a leading attributed-text prefix should still match the clean outbound cached text and be dropped as an echo.
• Why this is the smallest reliable guardrail: it exercises the exact normalization seam and the self-chat inbound decision seam without needing a live iMessage environment.
• Existing test that already covers this (if any): existing echo-cache and self-chat tests covered clean text and control-character-ish cases, but not the attributed prefix byte shape from iMessage echo detection fails for messages with attributed text prefix bytes #64309.
• If no new test is added, why not: N/A

User-visible / Behavior Changes

• iMessage reflected replies with attributed-text prefix bytes should stop re-entering the session as fresh inbound messages.

Diagram (if applicable)

Before:
[agent reply] -> [iMessage reflection with prefix bytes] -> [text mismatch in echo cache] -> [loop / duplicate reply]

After:
[agent reply] -> [iMessage reflection with prefix bytes] -> [prefix stripped before echo match] -> [echo dropped]

Security Impact (required)

• New permissions/capabilities? (No)
• Secrets/tokens handling changed? (No)
• New/changed network calls? (No)
• Command/tool execution surface changed? (No)
• Data access scope changed? (No)
• If any Yes, explain risk + mitigation:

Repro + Verification

Environment

• OS: macOS
• Runtime/container: local Node/pnpm checkout
• Model/provider: N/A
• Integration/channel (if any): iMessage
• Relevant config (redacted): iMessage echo-cache/self-chat test paths; local imsg access to Messages database

Steps

1. Cache a clean outbound iMessage reply text.
2. Simulate a reflected inbound message carrying a leading attributed-text prefix like \x93b\x02.
3. Verify the reflected text still matches the cached outbound text and is dropped as an echo.
4. Manually send live verification messages through imsg to a real iMessage target and confirm delivery on the recipient device.

Expected

• The prefixed reflected text normalizes to the clean outbound text and is treated as an echo.
• Live iMessage test messages deliver successfully to the target device.

Actual

• Before this patch, the prefixed reflected text missed echo matching and could re-enter the session.
• After this patch, the prefixed text matches in the targeted regression path, and the live verification sends delivered successfully.

Evidence

Attach at least one:

• Failing test/log before + passing after
• Trace/log snippets
• Screenshot/recording
• Perf numbers (if relevant)

Human Verification (required)

What you personally verified (not just CI), and how:

• Verified scenarios: targeted echo-cache regression test, targeted self-chat inbound regression test, successful branch commit hook (pnpm check), and live iMessage test sends via imsg during validation.
• Edge cases checked: prefixed reflected text with a GUID-backed outbound cache entry; self-chat is_from_me=true reflection path with the same prefix shape; real iMessage delivery path remained functional while validating on macOS.
• What you did not verify: a full live OpenClaw+iMessage end-to-end loop reproduction with the branch wired into an enabled iMessage channel.

Review Conversations

• I replied to or resolved every bot review conversation I addressed in this PR.
• I left unresolved only the conversations that still need reviewer or maintainer judgment.

Compatibility / Migration

• Backward compatible? (Yes)
• Config/env changes? (No)
• Migration needed? (No)
• If yes, exact upgrade steps:

Risks and Mitigations

• Risk: the prefix stripper could normalize an unusual legitimate leading control-byte sequence that was not an iMessage attributed-text marker.
  • Mitigation: the logic is intentionally narrow, only strips short leading marker-like runs, and is covered by targeted tests instead of broad text rewriting.