[recipes] Google Activity import: include Gemini answers from safeHtmlItem

[araece]

[recipes] Google Activity import: include Gemini's answers from safeHtmlItem

What

The Google Activity importer currently reads only title and time from Takeout's
MyActivity.json records — which for Gemini Apps means it imports what the user asked,
but never what Gemini answered.

It turns out the answers ARE in the export: most Gemini prompt records carry the response
as HTML in a top-level safeHtmlItem field:

{
  "title": "Prompted How does polarization work in sunglasses?",
  "time": "2026-03-02T09:15:00.000Z",
  "safeHtmlItem": [{ "html": "<p>Polarized lenses contain a <strong>filter</strong> ..." }]
}

Verified against a real export (July 2026): 1,161 of 1,640 Gemini records included it
(median ~420 chars, p90 ~3.7k, max ~86k). The widespread "Takeout exports prompts only"
belief appears to be outdated.

Changes

• extractGeminiAnswer() + dependency-free htmlToText() — strips the safeHtmlItem
  HTML to plain text (tags removed, entities decoded, <li> → - bullets), capped at
  1,000 chars per answer.
• groupByDay() appends Gemini's answer: … beneath the prompt title. Records without
  safeHtmlItem — all other categories, and the ~30% of Gemini records lacking it — behave
  exactly as before (the helper is a no-op for them).
• Summarization prompt: one added paragraph telling the model to use answers to capture
  what was learned/decided, but to attribute facts that appear only in an answer as
  AI-generated ("Gemini answered/suggested …") rather than established truth.
• Transcript cap raised 6k → 12k chars for Gemini Apps only, so answers survive the
  truncation (~3.4k tokens ≈ $0.0005/day at gpt-4o-mini rates). Other categories keep 6k.
• README: documents the discovery.

Why it matters

Day summaries built from prompts alone can only say what the user was asking about.
With the answers included, they can capture what the user actually found out — which is
the part worth remembering in a personal knowledge base.

Requirements

Unchanged from the existing recipe: a working Open Brain setup, an OpenRouter API key,
Node.js 18+. No new services, dependencies, or secrets (metadata.json bumped to 1.1.0,
gemini tag and description updated).

Testing

Tested against my own Open Brain instance:

• node --check passes.
• Dry-run (--dry-run --raw --verbose --categories "Gemini Apps") against synthetic
  records: answer extracted and appended, HTML/entities cleaned, answerless record
  untouched, other categories unaffected.
• The same parsing approach is running in production on my instance against a real
  1,640-record export.

Not in scope (possible follow-up)

Localized exports use different filenames/prefixes (German: MeineAktivitäten.json,
titles starting Eingegebener Prompt: ), which findMyActivityFiles()/the filters don't
match — non-English Takeout exports are currently invisible to this recipe. Happy to send
that separately if there's interest.

🤖 Generated with Claude Code

[github-actions]

OB1 PR Gate

✅ Folder structure — All files are in allowed directories
✅ Required files — README.md and metadata.json found in all contribution folders
✅ Metadata valid — All metadata.json files passed JSON Schema validation
✅ No credentials — No API keys, tokens, or secrets detected
✅ SQL safety — No destructive SQL or core table modifications
✅ Category artifacts — Required file types present for each category
✅ PR format — Title follows [category] Description format
✅ No binary blobs — No oversized or binary files
✅ README completeness — All READMEs include Prerequisites, Steps, and Expected Outcome
✅ Contribution dependencies — All declared skill and primitive dependencies exist and are linked in README
✅ LLM clarity review — Covered by Claude PR Review workflow
✅ Remote MCP pattern — No local MCP server patterns detected — uses remote MCP correctly
✅ Tool audit link — Extensions/integrations link to the MCP Tool Audit guide
✅ Scope check — All changes are within the contribution folder(s)
✅ Internal links — All relative links in READMEs resolve to existing files

Result: All 15 checks passed! Ready for human review.

---

Post-Merge Tasks

These don't block merge — they're reminders for admins after this PR lands.

• Add Google Activity Import to recipes/README.md
• Add Alan Shurafa (@alanshurafa) to CONTRIBUTORS.md
• Post in OB1 Discord #show-and-tell