Best possible solution:

Keep the stack PR open, make public identity reads non-mutating and malformed-row tolerant, then land only after maintainer API approval, branch refresh, and redacted real plugin/runtime proof.

Do we have a high-confidence way to reproduce the issue?

Yes. Source inspection shows the public reader delegates to a parse-all loader, and the public identity helper delegates to a target resolver whose callee can persist session metadata; I did not run tests because this was a read-only cleanup review.

Is this the best way to solve the issue?

No, not as a final landing shape. The identity boundary is plausible for the SQLite migration, but the public SDK contract should use read-only identity resolution, tolerate malformed transcript rows, and have maintainer API approval plus real proof before publishing.

Full review comments:

• [P2] Keep identity resolution read-only — src/plugin-sdk/session-transcript-runtime.ts:109
resolveSessionTranscriptIdentity calls resolveSessionTranscriptRuntimeTarget, which can route through resolveAndPersistSessionFile and write sessionFile/updatedAt when a row lacks the current transcript path. A public identity read should not change session activity or create file-backed metadata; derive the identity read-only or use a read-only target resolver.
  Confidence: 0.85
• [P2] Skip malformed transcript lines in public reads — src/plugin-sdk/session-transcript-runtime.ts:125
readSessionTranscriptEvents exposes loadTranscriptEvents through a new public SDK subpath, but that loader parses every streamed JSONL line without catching failures. Existing transcript readers tolerate malformed rows, so one bad row can prevent plugins from reading later valid session history unless this public reader catches parse failures and continues.
  Confidence: 0.88

Overall correctness: patch is incorrect
Overall confidence: 0.89

AGENTS.md: found and applied where relevant.

Codex review notes: model internal, reasoning high; reviewed against 86b24ac2b2fc.

Label justifications:

• P2: This is a normal-priority public SDK/session-transcript migration PR with compatibility impact but no standalone current production outage.
• merge-risk: 🚨 compatibility: The diff adds a documented public plugin SDK subpath and package export that third-party plugins may adopt after release.
• merge-risk: 🚨 session-state: The proposed identity/read path can touch session metadata and can reject recoverable transcript history until fixed.
• rating: 🧂 unranked krab: Overall readiness is 🧂 unranked krab; proof is 🧂 unranked krab and patch quality is 🦪 silver shellfish.
• status: 📣 needs proof: The PR needs real behavior proof before ClawSweeper can clear the contributor ask. Needs real behavior proof before merge: The PR body lists tests and implementation-thread checks but no redacted real plugin/runtime output, terminal transcript, logs, or artifact showing transcript identity resolution and event reads after the patch; add proof to the PR body so a fresh ClawSweeper review can re-run, or ask a maintainer to comment @clawsweeper re-review if it does not trigger automatically. After adding proof, update the PR body; ClawSweeper should re-review automatically. If it does not, the PR author or someone with repository write access can comment @clawsweeper re-review.

PR surface:

Source +825, Tests +579, Docs +3, Config +4, Generated 0, Other +4. Total +1415 across 20 files.

What I checked:

• AGENTS policy applied: The plugin SDK guide treats this directory as the public contract between plugins and core, so adding a subpath is compatibility-sensitive review scope. (src/plugin-sdk/AGENTS.md:1, 86b24ac2b2fc)
• Current main lacks the new public subpath: Current main lists session-store-runtime and session-transcript-hit, but no session-transcript-runtime package export or SDK entrypoint, so the central change is not already implemented. (package.json:982, 86b24ac2b2fc)
• Latest release lacks the new public subpath: The latest release tag has session-store-runtime and session-transcript-hit, but not session-transcript-runtime, so this is not shipped behavior either. (package.json:957, 844f405ac1be)
• PR adds a public package export: The PR head adds ./plugin-sdk/session-transcript-runtime to package.json, making this a public third-party plugin import surface once released. (package.json:952, f6d594959907)
• Identity helper calls the runtime target resolver: The new public resolveSessionTranscriptIdentity delegates to resolveSessionTranscriptRuntimeTarget, even though the public API only returns identity fields. (src/plugin-sdk/session-transcript-runtime.ts:109, f6d594959907)
• Runtime target resolution can persist session metadata: The resolver path can call resolveAndPersistSessionFile, which writes sessionFile and updatedAt to the session store when the target is missing or changed. (src/config/sessions/session-file.ts:46, 86b24ac2b2fc)

Likely related people:

• jalehman: Authored this PR, the linked Path 3 tracker, and the merged runtime transcript identity/state prerequisite that this public SDK layer builds on. (role: active migration stack owner; confidence: high; commits: f6d594959907, 57aa52c1e33c, efa9a6110b4e; files: src/plugin-sdk/session-transcript-runtime.ts, src/config/sessions/session-accessor.ts, src/agents/embedded-agent-runner/transcript-runtime-state.ts)
• jack-stormentswe: Authored the transcript streaming change that explicitly preserved malformed-line tolerance in the sibling readers this public helper should match. (role: original streaming helper contributor; confidence: medium; commits: 9654239af9f9; files: src/config/sessions/transcript-stream.ts, src/config/sessions/transcript.ts, src/gateway/session-compaction-checkpoints.ts)
• steipete: Merged the transcript streaming/tolerance change and recently touched plugin SDK runtime documentation and transcript helper docs in the same area. (role: recent adjacent contributor and merger; confidence: medium; commits: 9654239af9f9, c74fd6f01540, feffb6d02ff8; files: src/config/sessions/transcript-stream.ts, src/plugin-sdk/session-transcript-hit.ts, docs/plugins/sdk-runtime.md)

• 🦀 challenger crab: rare, exceptional readiness with strong proof, clean implementation, and convincing validation.
• 🦞 diamond lobster: very strong readiness with only minor maintainer review expected.
• 🐚 platinum hermit: good normal PR, likely mergeable with ordinary maintainer review.
• 🦐 gold shrimp: useful signal, but proof or patch confidence is still limited.
• 🦪 silver shellfish: thin signal; proof, validation, or implementation needs work.
• 🧂 unranked krab: not merge-ready because proof is missing/unusable or there are serious correctness or safety concerns.
• 🌊 off-meta tidepool: rating does not apply to this item.

Shiny media proof means a screenshot, video, or linked artifact directly shows the changed behavior. Runtime, network, CSP, and security claims still need visible diagnostics.

• ClawSweeper keeps one durable marker-backed review comment per issue or PR.
• Re-runs edit this comment so the latest verdict, findings, and automation markers stay together instead of adding duplicate bot comments.
• A fresh review can be triggered by eligible @clawsweeper re-review comments, exact-item GitHub events, scheduled/background review runs, or manual workflow dispatch.
• PR/issue authors and users with repository write access can comment @clawsweeper re-review or @clawsweeper re-run on an open PR or issue to request a fresh review only.
• Maintainers can also comment @clawsweeper review to request a fresh review only.
• Fresh-review commands do not start repair, autofix, rebase, CI repair, or automerge.
• Maintainer-only repair and merge flows require explicit commands such as @clawsweeper autofix, @clawsweeper automerge, @clawsweeper fix ci, or @clawsweeper address review.
• Maintainers can comment @clawsweeper explain to ask for more context, or @clawsweeper stop to stop active automation.