メニュー
Sovereign web search, scrape, and source preservation (SearXNG + Crawl4AI) for Mycroft and Spotlight workflows; Firecrawl is an optional fallback.
Codex または Claude でインストール この Prompt をコピーして Codex、Claude、または他のアシスタントに貼り付けると、Skill ページを確認してインストールできます。
SOC 職業分類に基づく
| name | web-acquisition |
| description | Sovereign web search, scrape, and source preservation (SearXNG + Crawl4AI) for Mycroft and Spotlight workflows; Firecrawl is an optional fallback. |
Mycroft acquires web sources sovereign by default — no API key, no vendor account:
python3 "$MYCROFT_DIR/tools/searxng-search.py" "<query>" (local SearXNG JSON API). Add --limit N, --categories news, or --time-range month.python3 "$MYCROFT_DIR/tools/scrape.py" <url> (Crawl4AI → clean markdown). Local PDFs: python3 "$MYCROFT_DIR/tools/scrape.py" <file.pdf> --pdf (pdftotext).python3 "$MYCROFT_DIR/tools/sitemap.py" <domain> (robots.txt / sitemap.xml enumeration).mycroft-fetch scrape <url> / mycroft-fetch search "<query>" — writes an evidence record with acquisition_method (crawl4ai / searxng / firecrawl), a SHA-256, and the access timestamp.Firecrawl is the optional escape hatch, reached only when FIRECRAWL_API_KEY is set and the sovereign tool can't do the job — a hard anti-bot target that defeats Crawl4AI (scrape), or an exhaustive search union. With no key, Mycroft runs pure-sovereign; the tools fall back to Firecrawl automatically when it is present.
sources/raw/.knowledge-workspace write flow with an explicit logical_space and idempotency key.For Mycroft durable knowledge:
sources/raw/sources/processed/wiki/For Spotlight casework:
cases/{project}/research/evidence/ or captures/Search, read, link, audit, and journal durable Mycroft or Spotlight knowledge through the storage-neutral Knowledge Workspace Port. Use for vault questions, durable knowledge writes, morning briefs, audits, Spotlight context lookup, approved Spotlight ingestion, OpenKnowledge health checks, legacy QMD/Obsidian migration, and any request that names a knowledge logical space.
Scoutpost integration for scouts, monitoring, and beat signals via the scout CLI or REST API.
Claim-to-evidence grounding discipline for Mycroft skills and recipes. Use when extracting findings, assigning confidence tags, fact-checking claims, ingesting vault notes, or deciding whether a piece of source material is a lead, a partially grounded claim, or a verifiable finding.
Ingest raw material, Spotlight handoffs, URLs, files, and text into the Mycroft knowledge vault.
Fact-check drafts, claims, and source assertions through Mycroft's SIFT recipe, escalating to Spotlight for deeper adversarial review when available. Uses the fact-check profile of epistemic-grounding — emits the full grounding object per the spec.
Eliminate AI-generated writing patterns that erode reader trust. Activate when writing articles, documentation, press releases, or any content where AI patterns would undermine credibility. For journalists using AI assistance who need human-sounding output.