fix: support gemini transcript parsing in summarize hook

[themactep]

This PR adds support for gemini transcript parsing in the summarize hook, detecting platform source and handling both JSON and JSONL formats with cross-platform role mappings.

[coderabbitai]

No actionable comments were generated in the recent review. 🎉

ℹ️ Recent review info

⚙️ Run configuration

Configuration used: Organization UI

Review profile: CHILL

Plan: Pro

Run ID: df9d720c-070f-49f3-83cf-ea2b18acfd13

📥 Commits

Reviewing files that changed from the base of the PR and between 189a19d and 0816fad.

📒 Files selected for processing (2)

• plugin/scripts/worker-service.cjs
• src/shared/transcript-parser.ts

🚧 Files skipped from review as they are similar to previous changes (1)

• src/shared/transcript-parser.ts

---

Summary by CodeRabbit

• New Features

  • Support for both JSON and line-delimited JSON transcript formats

• Improvements

  • Include platform source indicator in summary requests and completions
  • More robust message extraction across platform transcript shapes and role/type variants
  • Skip empty/missing message content and concatenate multi-part text reliably
  • Improved error logging and parsing fallback behavior

Walkthrough

The summarize handler now derives and includes a platform source ('gemini-cli' or 'claude') from hook metadata in its API payloads. The transcript parser now accepts JSON objects/arrays or JSONL, improves role/type matching, and strengthens content extraction and error handling.

Changes

Cohort / File(s)	Summary
Platform Source Tracking src/cli/handlers/summarize.ts	Derives platformSource from metadata?.hook_event_name ('gemini-cli' if present, else 'claude') and includes it in /api/sessions/summarize and /api/sessions/complete payloads; session completion call now sends contentSessionId plus platform_source/platformSource.
Transcript Parser Format & Robustness src/shared/transcript-parser.ts	Refactored extractLastMessage to parse either JSON (object/array with messages) or JSONL (fallback), improved role matching to platform-specific type values (`assistant

Estimated code review effort

🎯 3 (Moderate) | ⏱️ ~20 minutes

Poem

🐰 I hopped through lines both new and old,

Found JSON, JSONL, and stories told.
Gemini, Claude—now both in view,
Payloads carry the platform true.
Tiny paws applaud the clever glue.

🚥 Pre-merge checks | ✅ 3

✅ Passed checks (3 passed)

Check name	Status	Explanation
Title check	✅ Passed	The title accurately reflects the main change: adding support for Gemini transcript parsing in the summarize hook, which is the primary objective of this PR.
Description check	✅ Passed	The description is directly related to the changeset, covering platform source detection, JSON/JSONL format handling, and cross-platform role mappings mentioned in the file changes.
Docstring Coverage	✅ Passed	Docstring coverage is 100.00% which is sufficient. The required threshold is 80.00%.

_{✏️ Tip: You can configure your own custom pre-merge checks in the settings.}

✨ Finishing Touches

🧪 Generate unit tests (beta)

• Create PR with unit tests

---

Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out.

❤️ Share

• X
• Mastodon
• Reddit
• LinkedIn

_{Comment @coderabbitai help to get the list of available commands and usage tips.}

[coderabbitai]

Actionable comments posted: 1

🧹 Nitpick comments (1)

src/shared/transcript-parser.ts (1)

30-42: Avoid exception-driven format detection on the hot path.

Claude JSONL still pays for a full JSON.parse(...) failure before the real parse. A cheap first-character check would keep the common path out of exception handling.

♻️ Possible simplification

-  // Try parsing as standard JSON (Gemini CLI format)
-  try {
-    const data = JSON.parse(rawContent);
-    if (Array.isArray(data.messages)) {
-      messages = data.messages;
-    } else if (Array.isArray(data)) {
-      messages = data;
-    } else {
-      isJSONL = true;
-    }
-  } catch {
-    isJSONL = true;
-  }
+  // Try parsing as standard JSON (Gemini CLI format)
+  const firstChar = rawContent[0];
+  if (firstChar === '{' || firstChar === '[') {
+    try {
+      const data = JSON.parse(rawContent);
+      if (Array.isArray(data.messages)) {
+        messages = data.messages;
+      } else if (Array.isArray(data)) {
+        messages = data;
+      } else {
+        isJSONL = true;
+      }
+    } catch {
+      isJSONL = true;
+    }
+  } else {
+    isJSONL = true;
+  }

🤖 Prompt for AI Agents

Verify each finding against the current code and only fix it if needed.

In `@src/shared/transcript-parser.ts` around lines 30 - 42, The current code
always calls JSON.parse(rawContent) inside a try/catch which is expensive on the
hot path; replace that with a cheap first-character check on rawContent (e.g.,
const first = rawContent.trimStart()[0]) and only attempt JSON.parse when first
=== '[' or first === '{' (then run the existing JSON.parse block to populate
messages or mark isJSONL accordingly); otherwise set isJSONL = true immediately.
Update the logic around the JSON.parse call and variables rawContent, isJSONL,
and messages so you avoid throwing/handling exceptions for non-JSON transcripts.

🤖 Prompt for all review comments with AI agents

Verify each finding against the current code and only fix it if needed.

Inline comments:
In `@src/shared/transcript-parser.ts`:
- Around line 68-75: The loop over messages should skip non-object/nullable
entries before accessing msg.type; inside the for loop that iterates messages
(the block using foundMatchingRole and calling targetTypes.includes(msg.type)),
add a guard like if (typeof msg !== 'object' || msg === null) continue so
primitive/null entries are skipped, then proceed to use
targetTypes.includes(msg.type) and extract msg.message?.content or msg.content
safely (refer to variables/functions: messages, targetTypes, foundMatchingRole,
msgContent, msg.message?.content).

---

Nitpick comments:
In `@src/shared/transcript-parser.ts`:
- Around line 30-42: The current code always calls JSON.parse(rawContent) inside
a try/catch which is expensive on the hot path; replace that with a cheap
first-character check on rawContent (e.g., const first =
rawContent.trimStart()[0]) and only attempt JSON.parse when first === '[' or
first === '{' (then run the existing JSON.parse block to populate messages or
mark isJSONL accordingly); otherwise set isJSONL = true immediately. Update the
logic around the JSON.parse call and variables rawContent, isJSONL, and messages
so you avoid throwing/handling exceptions for non-JSON transcripts.

🪄 Autofix (Beta)

Fix all unresolved CodeRabbit comments on this PR:

• Push a commit to this branch (recommended)
• Create a new PR with the fixes

---

ℹ️ Review info

⚙️ Run configuration

Configuration used: Organization UI

Review profile: CHILL

Plan: Pro

Run ID: b522b3c7-2340-4d47-91ff-b77069ae4ee4

📥 Commits

Reviewing files that changed from the base of the PR and between cde4faa and 189a19d.

📒 Files selected for processing (5)

• plugin/scripts/mcp-server.cjs
• plugin/scripts/worker-service.cjs
• plugin/ui/viewer-bundle.js
• src/cli/handlers/summarize.ts
• src/shared/transcript-parser.ts

[thedotmack]

Closed during the April 2026 backlog cleanup. The underlying bug is now tracked in #1909, which is the single canonical issue being addressed by the maintainer. Thanks for taking the time to report — your symptoms and repro are captured in the consolidated ticket.