IMPACT by PressTitan · reference Scripts.
Scripts.
Deterministic work, no pip.
All under impact/scripts/. Python 3.10+. No pip deps.
Reference.24 tools.
Path: impact/scripts/ · returns JSON| Script | Command | Returns |
|---|---|---|
| init_state.py | --dir --domain | Vault skeleton |
| init_vault.py | --dir --name | memory, intake, kpis, pr/, snapshots/ |
| upgrade_state.py | --from-dir --to-dir [--apply] | Move leftover state into one vault |
| health_probe.py | URL [--site-checks] | Status, TTFB, title, H1 |
| batch_probe.py | URL [URL…] | Array of probes |
| watchlist_probe.py | --dir | Probe config.watchlist, save snapshots |
| snapshot_save.py | --dir | Append probe JSONL |
| snapshot_diff.py | --dir [--url] | Last two snapshots |
| fetch_text.py | example.com | robots, sitemaps, llms.txt |
| robots_ai.py | example.com | Googlebot + AI crawlers |
| sitemap_urls.py | SITEMAP_URL | First N loc |
| extract_links.py | URL | Internal vs external |
| find_schema.py | URL | JSON-LD types |
| parse_gsc.py | pages.csv | Low-CTR rows |
| gsc_ingest.py | pages.csv [--prior prior.csv] | Striking distance, CTR gap, decay, totals |
| expected_ctr.py | POSITION | Heuristic CTR |
| compare_titles.py | --current --candidate | Token diff |
| score_actions.py | candidates.json | Winner + ranked |
| claim_check.py | file.html --dir VAULT | Unsourced claims vs truth ledger |
| write_run_record.py | --out DIR | Markdown run record |
| write_ticket.py | --out --decision … | Six-field ticket |
| learn_from_log.py | VAULT | Lane weights |
| portfolio_scan.py | --parent DIR | One line per vault |
| indexnow_ping.py | --host --key --url [--dry-run] | IndexNow POST or payload |
24 scripts. A failing script prints JSON, not a stack trace — a traceback is a bug in the script.
The 33 skills that call them: skill catalog. How they fit the run: the manual.