Greptile Summary

This PR narrows the echo-cache text normalization in the iMessage extension to strip only leading imsg replacement/control-byte corruption characters (U+FFFD and C0/C1 control ranges, excluding TAB and LF), while preserving interior bytes so that multiline or otherwise mixed-content keys remain non-lossy. The change is limited to extensions/imessage/src/monitor/echo-cache.ts with matching test coverage and a CHANGELOG entry.

Confidence Score: 5/5

Safe to merge — narrowly scoped fix with good test coverage and no logic errors found.

The normalization ordering (replace → trim → strip leading corruption → trim) is correct, the for...of iteration correctly handles multi-code-unit characters, and both remember and has go through the same normalizeEchoTextKey path so keys are symmetric. No P0/P1 issues identified.

No files require special attention.

_{Reviews (1): Last reviewed commit: "fix(imessage): normalize leading echo-ca..." | Re-trigger Greptile}

Codex review: keeping this open for maintainer follow-up; there is still a little grit to resolve.

Keep this PR open. Current main and the latest release still normalize iMessage sent-message echo-cache text with only CRLF normalization plus trim, so leading U+FFFD/C0/C1 corruption still misses the cached outbound key. The PR is a focused implementation candidate carrying forward the closed unmerged #62191 work, with tests that preserve interior control/replacement bytes. It overlaps #63581 and #66169, so maintainers should consolidate or pick the canonical patch rather than close this as implemented.

Best possible solution:

Keep this PR open for maintainer review. The best path is to land this narrow iMessage plugin echo-cache normalization, or combine it with #63581/#66169 into one canonical patch that handles the known leading-prefix variants, preserves interior message bytes, keeps the focused regression tests, and validates with the targeted iMessage monitor tests plus the changed gate.

What I checked:

• current-main-normalization-gap: Current main's normalizeEchoTextKey() only does CRLF-to-LF normalization and trim; it does not strip leading U+FFFD or C0/C1 corruption before echo-cache comparison. (extensions/imessage/src/monitor/echo-cache.ts:26, 293348b429b4)
• current-main-test-gap: The focused echo-cache tests cover CRLF/trim, message-id matching, placeholder IDs, and TTL behavior, but current main has no leading replacement/control-prefix regression test or interior-byte preservation test. (extensions/imessage/src/monitor/monitor-provider.echo-cache.test.ts:9, 293348b429b4)
• latest-release-same-gap: The latest release tag v2026.4.26 resolves to be8c246, and its echo-cache implementation has the same CRLF-plus-trim normalizeEchoTextKey() behavior. (extensions/imessage/src/monitor/echo-cache.ts:26, be8c24633aaa)
• pr-diff-scope: The PR head adds isLeadingImsgEchoCorruptionChar() and stripLeadingImsgEchoCorruption() in the iMessage extension echo cache, then applies it inside normalizeEchoTextKey(). The change is scoped to CHANGELOG.md, echo-cache.ts, and the focused echo-cache test file. (extensions/imessage/src/monitor/echo-cache.ts, 80b1dd16bf00)
• pr-test-coverage: The PR adds regression coverage for leading U+FFFD/control-byte prefixes matching the clean outbound text, and for preserving line breaks, tabs, interior control bytes, and interior U+FFFD instead of collapsing message body text. (extensions/imessage/src/monitor/monitor-provider.echo-cache.test.ts, 80b1dd16bf00)
• source-pr-context: Source PR fix(imessage): strip U+FFFD garbage chars from echo text key normalization #62191 is closed unmerged and its comments show this PR was opened as a narrow replacement after the original branch could not be safely updated; the prior discussion also says TTL and destination_caller_id work had already moved upstream, leaving the text-key normalization as the remaining fix. (eec4bb724d08)

Likely related people:

• rmarr: GitHub path history shows b29e180 introduced the central iMessage echo-cache/self-chat dedupe behavior, including post-send cache population, text TTL changes, and targeted tests. (role: introduced echo-cache/self-chat dedupe behavior; confidence: high; commits: b29e180ef407; files: extensions/imessage/src/monitor/echo-cache.ts, extensions/imessage/src/monitor/deliver.ts, extensions/imessage/src/monitor/self-chat-dedupe.test.ts)
• Ayaan Zaidi [email protected]: Authored and co-authored multiple adjacent iMessage echo/self-chat fixes in current-main history, including trying all inbound echo ids and dropping ambiguous reflected self-chat echoes. (role: adjacent echo/self-chat maintainer; confidence: high; commits: 4a5885df3a36, e3e2a19ab7f1, 1ee4a1606e9a; files: extensions/imessage/src/monitor/inbound-processing.ts, extensions/imessage/src/monitor/self-chat-dedupe.test.ts)
• neeravmakwana: Authored adjacent merged iMessage fixes for destination_caller_id/self-chat classification and wrapped imsg RPC text fields, and opened earlier closed PR fix(imessage): catch delayed reflected echoes #59980 for delayed/corrupted echo normalization. (role: adjacent merged-feature contributor; confidence: medium; commits: 9267c3f8f2f2, 8ed7c95a6a82, 782b5622b6a5; files: extensions/imessage/src/monitor/inbound-processing.ts, extensions/imessage/src/monitor/types.ts, extensions/imessage/src/monitor/parse-notification.ts)
• vincentkoc: The source PR discussion says vincentkoc coordinated ProjectClownfish's narrow repair while preserving contributor credit; GitHub history also shows recent iMessage monitor retry/type-seam maintenance in the extension area. (role: recent maintainer and follow-up coordinator; confidence: medium; commits: 35a784c16563, ea71a59127a0, 503b43f43f61; files: extensions/imessage/src/monitor/monitor-provider.ts, extensions/imessage/src/monitor/watch-error-log.ts, extensions/imessage/src)

Remaining risk / open question:

• The PR overlaps active open PRs imessage: strip NUL bytes from echo-cache text normalization #63581 and [codex] fix(imessage): normalize attributed echo prefixes #66169; maintainers should consolidate or choose one canonical iMessage echo-cache prefix normalizer before landing competing branches.
• The normalization intentionally treats leading U+FFFD/C0/C1 bytes as echo-cache corruption within the existing 4s text fallback window, so there is a small functional false-positive risk if a real user sends the same text with such a leading prefix immediately after an outbound message.
• No supply-chain, workflow, dependency, lockfile, publishing, permission, or secret-handling risk was found in the PR diff.

Codex review notes: model gpt-5.5, reasoning high; reviewed against 293348b429b4.

Thanks for pushing this. We merged the narrower canonical fix in #73942: #73942, merge be445dd. It keeps normalization limited to observed leading attributedBody markers and leaves TTL and broader control-byte behavior separate. Testbox proof: tbx_01kqbp2m7s1mn1nd4fhgnp74wk focused iMessage tests plus check:changed passed. Closing this overlap.