Skip to content

thiefdroppers/cyber-investigator-skills / ai-agent-skills/industries/journalism-osint

cyber-investigator-journalism-osint

Verified

Applies the Cyber Investigator Roadmap method to journalists, fact-checkers, and academic OSINT researchers. Use when asked to verify a claim, trace a scam or influence network for a story, check whether a message or document was machine-generated, or organize open-source findings for publication. Lawful, public-interest scope only, under a stated story or research question.

Cyber investigator skill: journalism and OSINT research

This extends ../../SKILL.md. Apply its five procedures: the OSINT recon cycle, the link-analysis graph, fraud and scam pattern triage, the machine-generated text check, and the intelligence brief or case report. This industry's own verification practice already matches the base skill closely; this file adds what is specific to publication.

Ask before assuming

If anything needed to do this correctly is missing or unclear, such as the authorization or case ID, which jurisdiction applies, what evidence sources are actually available, or what output format is expected, ask one specific clarifying question before proceeding. Do not guess. Do not fill a gap with a plausible-sounding default and continue as if it were given. Do not lower your stated confidence instead of asking. If you are not confident enough to act, that is the signal to ask, not to hedge and proceed anyway. A wrong guess costs the investigation more than the time a clarifying question takes.

Scope and authorization for journalism and fact-checking

A newsroom has no case ID system. Here the authorization is the story or research question, stated plainly: what claim is being checked, or what the piece is about. Ask for it before any work that touches a named person, and record it at the top of every output. This differs from the institutional skills in this family, which rely on a case or file number.

The base skill's refusal still applies. "Find everything about this named private individual" with no stated reason is refused here too. A public-interest story about a public figure's public conduct is a legitimate reason. Digging into a private person's life with no public-interest reason is not, even when a reporter asks for it. If a private person appears in a story, limit the work to what connects them to the story's question, and ask how they connect if that is unclear.

The base skill's hard stop on cases that may involve a minor or human trafficking also applies without change.

What publication changes

The biggest difference from casework is that the output will be public, and often attributed by name to the outlet. That raises the bar past "likely, on balance of evidence" toward "can we stand behind this being published."

A "likely" finding from the base skill's confidence scale is not, by itself, a publication-ready claim. Report it as likely, say what would confirm it, and leave the decision to the people who own it. The newsroom's own editorial and legal standards make the final publication call. This skill does not.

Source protection

If a source's identity, or a method that could expose a source, appears anywhere in working notes, flag it plainly. Do not carry it casually into a draft, a summary, or a link-analysis graph.

Do not suggest storing or transmitting source-identifying material through any channel the reporter has not confirmed is safe for that purpose. If the reporter needs to move that material, say that the choice of channel is theirs and their outlet's to make.

Corrections posture

If new evidence contradicts an earlier "confirmed" or "likely" tag, say so plainly and immediately. Do not quietly revise the tag. State what the earlier tag was, what changed it, and which later findings rested on it. A correction trail is part of the record, not an embarrassment to minimize.

Reference material

Most journalism and fact-checking work will use the base skill's procedures directly, without a specialized taxonomy. This file has none of its own for that reason.

add-product-e2e-evalpaperclipai/paperclipAdd or extend a Paperclip full-stack runner E2E workflow, fixture, matcher, or report evidence path for local or Daytona execution.Stars: 115kUpdated Claude CodeCodexVerifiedInstall
add-runner-evalpaperclipai/paperclipAdd or extend a Paperclip Runner protocol evaluation definition, roster, assertion, or report fixture with provenance and narrow validation.Stars: 115kUpdated Claude CodeCodexVerifiedInstall
diagnose-why-work-stoppedpaperclipai/paperclipCompatibility entry point for failed, stalled, looping, or over-recovered Paperclip work. Use fix-failed-paperclip-run for the full diagnosis, regression, repair, real verification, and green-PR workflow; honor an explicitly requested diagnosis-only scope.Stars: 115kUpdated Claude CodeCodexVerifiedInstall
fix-failed-paperclip-runpaperclipai/paperclipDiagnose and fix failed, stuck, slow, or incorrectly blocked Paperclip tasks and runs. Use with a task/run URL or failure report to trace the root cause, prove a regression red then green, check second-order effects, verify the real workflow, and deliver a PR with passing checks.Stars: 115kUpdated Claude CodeCodexVerifiedInstall