Onboarding personalization, chat cold-start fix, Lexi chat UX, and backend hardening

[InfinityZero3000]

Summary

Two commits on top of the already-merged product-expansion work (#391):

7e152307 — feat(lexi-chat): onboarding personalization, cold-start latency fix, and chat UX improvements

• Persist onboarding goal/interest (previously silently dropped) and thread into ai-service's learner_profile via the existing internal batch-get pull + LearnerStateProfile.state_epoch cache-invalidation.
• Fix TraceCAG cache/KG version drift and add content-hash KG mutation detection.
• Fix a measured 169.5–478s cold-start chat latency: RetrievalServiceV3 (graph analytics + concept embeddings) and the JIT GLiNER model now warm as a non-blocking background task at boot instead of on whichever live request hits them first.
• Cap betweenness-centrality/community-detection cost on the real production KG size.
• Fix diagnose_node's broken Groq-fallback path and truncated-JSON max_tokens.
• Lexi chat: TTS defaults to the app-wide Sound setting until overridden; corrections highlight inline on the user's own message bubble; one-tap practice suggestion tied to the specific mistake; fixed a real data-loss bug where "already saved" vocabulary state reset after every app restart; new sessions can reference a short recap of the learner's last conversation.

cf90bdab — fix(backend): wallet cache consistency, streak race condition, SSRF hardening

• Wallet cache only invalidates when a transaction actually commits.
• claim_daily_reward now also invalidates the wallet cache.
• Row-level lock on streak read-modify-write to close a concurrency race.
• safe_http.resolve_pinned_ip: excludes multicast ranges that is_global alone let through (DNS-rebinding hardening gap).
• Scheduled Celery tasks for periodic news/YouTube/podcast content prefetch.
• Test coverage for PDF-based bulk-import endpoints.

Both commits were independently code-reviewed (3 separate reviews covering ai-service/backend-service/flutter-app) and every finding was fixed before commit.

Test plan

• Backend: full pytest suite (1363 passed, excluding the real-Postgres-only integration file which was run separately and passed)
• ai-service: pytest tests/trace_cag/ + related chat integration suites (404 passed)
• Flutter: flutter test across lexi_chat/vocabulary/youtube/books/core-services (175 passed)
• dart analyze / backend lint clean (only one pre-existing, unrelated info-level lint remains)
• CI on this PR

🤖 Generated with Claude Code

[vercel]

The latest updates on your projects. Learn more about Vercel for GitHub.

2 Skipped Deployments

Project	Deployment	Actions	Updated (UTC)
lexilingo-admin	Ignored		Aug 7, 2026 7:03am
lexilingo-app	Ignored		Aug 7, 2026 7:03am

[github-actions]

❌ PR-Agent failed to apply 'local' repo settings

The configuration file needs to be a valid TOML, please fix it.

---

Error message:
Cannot declare ('config',) twice (at line 61, column 8)

Configuration content:

[config]
# Use Google Gemini API
model = "google/gemini-2.5-pro"
fallback_models = ["google/gemini-2.5-flash"]

[pr_reviewer]
# Enable/disable features
require_focused_review = true
require_score_review = true
require_tests_review = true
require_security_review = true
require_estimate_effort_to_review = true

# Review settings
num_code_suggestions = 5
inline_code_comments = true
auto_review = true
remove_previous_review_comment = true

# Language and tone
extra_instructions = """
- Focus on code quality, security, and best practices
- Check for proper error handling
- Verify Clean Architecture principles
- Ensure proper test coverage
- Review API design and RESTful conventions
- Check for potential performance issues
- Verify proper logging and monitoring
"""

[pr_code_suggestions]
# Code improvement suggestions
num_code_suggestions = 5
extra_instructions = """
- Suggest improvements for code readability
- Recommend better naming conventions
- Propose performance optimizations
- Suggest security enhancements
"""

[pr_description]
# Auto-generate PR description
publish_description = true
add_original_user_description = true
extra_instructions = """
- Summarize main changes
- List affected components
- Note breaking changes
- Include testing notes
"""

[pr_questions]
# Auto-generate clarifying questions
num_questions = 3
extra_instructions = """
- Ask about design decisions
- Question potential edge cases
- Clarify implementation choices
"""

[config]
# Model configuration (uncomment to customize)
# model = "gpt-4-turbo-preview"
# model = "anthropic/claude-3-5-sonnet-20241022"
# fallback_models = ["gpt-3.5-turbo"]

# Git provider
git_provider = "github"

# Review triggers
pr_commands = [
    "/review",
    "/describe",
    "/improve",
    "/ask",
    "/update_changelog",
]