← ClaudeAtlas

cross-scanlisted

Cross-reference outputs from momentum-scan, base-breakout-scan, mean-reversion-scan, and unusual-options-scan to find tickers appearing in 2+ scans — the highest-conviction "agreement" picks where technical + flow signals stack. Use when the user wants overlap / consensus picks across the sister scans, a high-conviction daily watchlist, or to act on the four scans together rather than separately. Triggers on "cross-scan", "consensus picks", "overlap", "agreement across scans", "what's in multiple scans", "combine/merge scan outputs". Do NOT use for single-scan re-runs, single-ticker lookups, or fundamentals questions — invoke the relevant individual scan instead.
mthli/skills · ★ 2 · AI & Automation · score 69
Install: claude install-skill mthli/skills
# cross-scan Aggregate the latest snapshots from the four sister scans — **momentum-scan**, **base-breakout-scan**, **mean-reversion-scan**, **unusual-options-scan** — and surface tickers appearing in **two or more** of them. The premise: each scan answers a different question (what's running / what's setting up / what's oversold but fine / where is options flow positioning), and any single signal can be a fluke or a noise pattern. When **two or three** of these independent signals fire on the same name on the same day, that's the small subset most worth a research dig. By default this skill is purely an aggregator — it reads the four scans' existing state files. Pass `--refresh` to have it re-run any stale sister scan first (snapshots older than 3 days, including missing ones), so you can drive a "make sure everything is fresh and show me the overlap" workflow in one command. Refreshes run silently (no progress output) for ~1-3 minutes each. After each refresh cross-scan verifies that the snapshot date actually advanced — if a scan exited cleanly but didn't write new data (e.g. weekend save-skip), you get an explicit warning rather than a misleading ✓. Either way, refresh failures don't abort the report; the freshness header on the markdown output always tells you what you actually got. **Dependencies**: Python ≥ 3.10 standard library only for the aggregation itself. `--refresh` shells out to the sister scans, which need `uv` in `PATH` (they auto-fetch `yfinance`, `pandas