{"id":17343,"plugin_id":"plugins_6a7624ebb8648191918bc29197f7427e","kind":"skill","collection_source":null,"comparison_source":null,"observed_at":"2026-09-30T23:14:06.876Z","digest":"7c5bef758df63d940d392052425f3c84d19dbeba5214d3b932b96d17cc6c110d","against":null,"payload":{"description":"Plan and document lawful, ethical, reproducible open-source intelligence collection. Use for public web, social, corporate, media, mapping, archive, domain, or other open sources where provenance and verification matter.","included_files":[{"relative_path":"agents/openai.yaml","size_in_bytes":271}],"name":"osint-collector","skill_md_contents":"---\nname: osint-collector\ndescription: >-\n  Plan and document lawful, ethical, reproducible open-source intelligence\n  collection. Use for public web, social, corporate, media, mapping, archive,\n  domain, or other open sources where provenance and verification matter.\n---\n\n# OSINT Collector\n\nCollect only what the authorised objective requires. Public visibility does not\nitself establish permission to collect, retain, republish, or act on information.\n\n## Scope and safety\n\nObtain the questions, jurisdiction, legal and ethical constraints, identifiers,\ndate cutoff, geography, languages, risk, minimisation rule, delivery format, and\nstop conditions. Agree whether interaction, pretext, automation, archives, or\npaid sources are authorised; default to passive collection only.\n\n## Collection method\n\n1. Convert questions into a query and source matrix with alternative names,\n   transliterations, dates, entities, locations, and disconfirming searches.\n2. Plan safety for accounts, devices, links, downloads, malware, exposure,\n   subject contact, and investigator identity.\n3. Record each URL or stable ID, publisher, UTC timestamp, access path, visible\n   context, query, collector, and method.\n4. Preserve screenshots and native downloads where lawful; hash captured files\n   and retain relevant page, header, archive, and platform metadata.\n5. Distinguish live source, platform copy, archive, quotation, syndication, and\n   repost; trace significant claims toward the earliest available source.\n6. Verify identity with independent attributes. Keep name matches, handles,\n   accounts, organisations, and people separate until linked.\n7. Corroborate geolocation, chronolocation, ownership, authorship, and events;\n   state confidence and alternatives.\n8. Build a claim-to-source table and assess origin, proximity, incentives,\n   consistency, corroboration, manipulation risk, and gaps.\n9. Preserve changed or deleted content only lawfully and distinguish collection,\n   event, and publication times.\n10. Log queries, decisions, exclusions, and safety or legal stops reproducibly.\n\n## Output\n\nProvide a collection plan, query log, source register, capture and hash manifest,\nfindings matrix, confidence table, gaps, and legal, ethical, and safety notes.\n\n## Guardrails\n\nDo not hack, evade controls, abuse credentials, stalk, dox, deceive, contact\nsubjects, or access illicit material without explicit lawful authority and\nspecialist controls. Minimise sensitive data. Never identify a person or allege\nmisconduct from one source, resemblance, or automated match alone.\n"},"changes":[],"summary":"First saved snapshot. No earlier version is available for comparison.","summary_kind":"deterministic","summary_metadata":{}}