Tony_tech/athena-oracle
3
0
Fork 0
Code Issues 6 Pull Requests 2 Actions Packages Projects Releases Wiki Activity
34 Commits 5 Branches 1 Tag
ed6b7475dcc86576eaa0015815db597b83ee1236
Commit Graph

9 Commits

Author SHA1 Message Date
Epictetus ed6b7475dc Sprint1: pure-rule bucket classifier + scorer (taste before intelligence) 
- athena/scoring.py: deterministic bucket taxonomy (SHIPPING, LOCAL AI,
  PROBLEM SOLVED, MODEL RELEASE, RESEARCH, BUSINESS, INFRASTRUCTURE,
  CULTURE, UNCATEGORIZED) + component scores (shipping/utility/
  replication/enthusiast/novelty) with hype_penalty. No embeddings/LLM.
- Idempotent schema migration: 12 new columns incl actionability_score (reserved).
- attach_scoring() wired into pipeline.py AFTER store_entries, BEFORE render.
- Backfilled all existing rows; verified at box: null->scored via live run.
- Review report exposes every fired rule (editorial proof, not accuracy metric).
- Human-review tally (Published/Rejected/Borderline) is manual only.

Decision: founder directive 2026-07-15 — discover taxonomy before adding
intelligence. Sprint 2 (Lens) blocked until manual review completes.
 2026-07-15 04:18:21 +00:00
Epictetus 424da91364 FIX: first_seen re-stamp bug — persist true source publish date, idempotent upsert, backfill DB 2026-07-13 22:16:59 +00:00
Epictetus feba2fe7b7 feat: apply fix/adapter-health-1-2-9 changes (unified retry, failure_class, RSS enabled) 
- adapters/__init__.py: add http_get() unified retry helper + AdapterHTTPError
  + failure_class classification (429/5xx/4xx/error/zero_fetch/ok)
  + last_failure_class on SourceAdapter for pipeline capture
- pipeline.py: ENABLED_SOURCES now includes RSS
  + failure_class rollup in run_log (most severe across all adapters)
  + per-source failure_class in source_stats
- schema.sql: add failure_class column to run_log table

Backport of fix/adapter-health-1-2-9 branch (issues #1, #2, #9).
 2026-07-12 04:51:24 +00:00
Epictetus 7ee1af3d7b Disable RSS adapter from ENABLED_SOURCES (back to 5 approved sources); keep code for future use 2026-07-08 18:03:37 +00:00
Epictetus 2c6701f5a3 Add RSS Feeds adapter (6th source) — commercial AI news from 10 feeds 
Bug fix: date parsing was ISO-only (RFC 2822 feeds filtered out).
Added email.utils.parsedate_to_datetime() fallback for RSS dates.

10 feeds: TechCrunch AI, VentureBeat AI, The Verge AI, AI News,
The Decoder, MIT Tech Review AI, OpenAI Blog, Anthropic, Google AI, Meta AI.

3 dead feeds (Anthropic/Google/Meta 404), 7 working.
Score: authority (primary blogs 1.5x, industry 1.3x) × recency decay (48h half-life)
Age cutoff: 14 days. Score type: estimated.

Wired into pipeline.py ENABLED_SOURCES. Added RSS URL verification.
3/3 spot-checks passed (all URLs reachable, HTTP 200).

Commit: 62031be→c24a89f
 2026-07-08 15:27:59 +00:00
Epictetus 4ba270c166 Add Hugging Face adapter — model releases & adoption signal 
- adapters/huggingface.py: HF API adapter (models + datasets)
- Dual fetch: sort=likes (popularity) + sort=lastModified (fresh)
- AI relevance filter: pipeline_tag, library_name, tag matching
- Score: adoption (likes/downloads log-scale) + relevance (pipeline/library/tags)
- score_type: actual (real likes/downloads from HF API)
- Cross-source signal: GLM-5.2 top on both HN and HF
- Wired into ENABLED_SOURCES + verification in pipeline.py
- Live verify: DB likes match live API exactly (GLM-5.2: 3607, DeepSeek-R1: 13448)
- 99 total entries across 5 sources, all pipeline green
 2026-07-08 05:54:27 +00:00
Epictetus 6396b66b45 Add Hacker News adapter + wire into pipeline 
- adapters/hackernews.py: HN Firebase API adapter with AI keyword filtering
- Word-boundary matching to avoid substring traps (Britain/Guinea)
- Score: log(points) + log(comments), actual HN scores
- Wired into ENABLED_SOURCES + verification in pipeline.py
- Live test: 10 AI/ML stories fetched, all clean
 2026-07-08 05:32:28 +00:00
Epictetus b51be6ed48 Phase 5: cron entry script, soft-cap archive, run_log zero-fetch degradation 
- oracle-pipeline.sh: single cron entry point (pipeline -> summarize -> archive)
- archive.py: soft-cap archival to entries_archive (preserve, not delete)
- pipeline.py: record zero-fetch (rate-limited) runs as degraded in run_log.notes
Verified: full script runs exit 0, run_log captures per-source status.
 2026-07-08 04:07:54 +00:00
Epictetus 67c002b665 Initial commit: Oracle AI research pipeline (adapters, pipeline, summarize, query) 
Source-controlled baseline before Phase 5 cron. Excludes oracle.db,
logs/, and __pycache__ via .gitignore. Pipeline verified running
clean end-to-end (run_log write confirmed before conn.close()).
 2026-07-08 04:03:36 +00:00