User reported that the old prompt could emit 'sell' on backward-looking
capitulation ('Kevin sold after a 20% drop') — exactly the false signal
to avoid. v2 reframes every per-ticker field as forward-looking and
adds an explicit expected_move enum for the trading bot to weight.
Changes:
- New ExpectedMove enum (up_strong/up_mild/sideways/down_mild/
down_strong/unknown) in shared/schemas + shared/models, with
matching kevin_expected_move Postgres enum + column on
kevin_stock_mentions (migration e5f6a7b8c9d0). NOT NULL with
server_default 'unknown' so existing rows backfill cleanly.
- SYSTEM_PROMPT rewritten: action semantics now require a FORWARD
view; reactive sells get downgraded to 'watch' or skipped; the
rationale_quote must contain forward reasoning. Quality
checklist updated.
- _ANALYSIS_TOOL JSON schema gains expected_move (required).
- prompt_version v1 → v2 in config + infra + ad-hoc CLI default.
- pipeline.py persists ticker.expected_move into the new column.
Migration safety: the column is NOT NULL DEFAULT 'unknown' so 96
existing mentions auto-fill with 'unknown' (no forward call known
for backward analyses) without breaking any reads.
Cost to backfill the 27 existing analyses with v2 prompt: ~$3 LLM
spend. A follow-up reanalyze script will replay them after this
ships.
First production run hit Anthropic's per-account rate_limit_error (429) trying
to burn through 16 backfill videos in seconds. The SDK's built-in retry can't
recover because the rate limit window resets slower than the 3 retry attempts.
Added meet_kevin_inter_video_sleep_seconds (default 30s) to PipelineDeps and
main's _process_pending_videos loop. 16 backfill videos now take ~8 min (16 * 30s
sleeps + ~30s per LLM call) instead of bursting into the rate limit.
Previous refactor (89f01ad) moved to OpenRouter because no sk-ant-api-* key
was found in Vault. Turns out claude-agent-service-spare-{1,2} hold
sk-ant-oat01-* OAuth tokens (108 chars, scope user:inference, 1-year TTL,
minted via 'claude setup-token' — see memory id=832).
These tokens work with the Anthropic SDK via the auth_token= constructor
argument (routes to Authorization: Bearer ... instead of x-api-key: ...).
They consume the Enterprise Claude subscription quota rather than
per-call billing, so the OpenRouter zero-credit problem goes away.
- llm_analyzer.py: revert OpenAI client to AsyncAnthropic; tool-use API
+ cache_control restored
- config.py: openrouter_api_key -> anthropic_oauth_token; model slug
reverted from anthropic/claude-sonnet-4.5 -> claude-sonnet-4-5
- main.py: AsyncOpenAI -> AsyncAnthropic(auth_token=...), drop OpenRouter
attribution headers
- pyproject: openai>=1.50 -> anthropic>=0.40 in meet_kevin extras
- tests: mocks ported back to messages.create + tool_use blocks
User's Vault has openrouter_api_key but no direct sk-ant-* Anthropic key.
OpenRouter passes through Claude Sonnet 4.6 (~3% markup over Anthropic
list pricing) and matches the existing gpt_mini_endpoint pattern used
by recruiter-responder.
- Replace anthropic.AsyncAnthropic with openai.AsyncOpenAI + base_url
- Convert Anthropic tool-use API to OpenAI function-calling
- System prompt unchanged (analyst instructions are model-agnostic)
- Drop cache_control (not in OpenAI API); revisit later if cost matters
- Model slug: anthropic/claude-sonnet-4.5 (OpenRouter's current Claude tier)
- Pricing: $3.10/M input, $15.50/M output (OpenRouter pass-through)
- Config field anthropic_api_key -> openrouter_api_key
- pyproject extras: anthropic>=0.40 -> openai>=1.50
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
- Implement CaptionResult frozen dataclass for structured caption data
- Add parse_srt() to parse SubRip format with flexible timestamp handling
- Add extract_captions() async function using yt-dlp subprocess wrapper
- Prefer manual captions over auto-generated; clean up SRT files after parsing
- Add 16 comprehensive tests covering edge cases (empty input, malformed SRT,
timestamp variations, language extraction, manual vs auto selection)
- Type-safe implementation with full mypy --strict compliance
- Add sample.srt fixture with 3 segments mentioning NVDA for test reference
