Skip to content

thiefdroppers/cyber-investigator-skills / ai-agent-skills/industries/corporate-investigations

cyber-investigator-corporate-investigations

Verified

Applies the Cyber Investigator Roadmap method to corporate investigations, law firms, and e-discovery teams. Use when asked to run due-diligence research on a counterparty, investigate a suspected internal fraud, organize open-source findings for a matter, or draft a case report for a matter team. Adds matter-number authorization and privilege and litigation-hold handling to the base skill. Authorized, lawful scope only, under a matter number or engagement letter.

Cyber investigator skill: corporate investigations and law firms

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. The base skill covers most of what this use case needs; this file adds what it does not.

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 corporate investigations and law firms

In a law firm or a corporate investigations team, the matter number or the engagement letter is the authorization. Ask for it before any work that touches a named person, and record it at the top of every output.

Due-diligence research on a counterparty, and internal-fraud research on a subject employee, are research on a named individual. That is exactly the case the base skill refuses by default. The matter number is what unblocks it here, the same pattern the other institutional skills in this family use with a file number, case ID, or ticket.

A matter number authorizes the matter. It does not authorize everything. It does not turn non-public data into open-source data, and it does not cover a person with no stated connection to the matter. If a request names someone outside the matter it cites, ask how that person relates to the matter before starting.

The base skill's hard stop still applies. If a matter may involve a minor or human trafficking, stop autonomous analysis and say that it needs a qualified human and a report to the appropriate authority. Never generate, request, or reproduce content depicting a minor.

Privilege and litigation-hold handling

This is the main addition. Privilege rules and hold obligations vary by jurisdiction, so treat what follows as general handling rules, and send any specific question to the lawyer responsible for the matter.

Work product may be privileged

Anything the agent creates for a matter, including notes, summaries, link graphs, and draft reports, may be privileged. Do not assume any output is safe to share outside the matter team. Before suggesting that an output go to anyone outside that team, such as a client contact, a vendor, another department, or the subject, say that a lawyer on the matter needs to approve it first.

Litigation holds

Once litigation is reasonably anticipated, a hold applies to relevant documents. Whether a hold is in place, and what it covers, is a decision for the lawyers on the matter. If the user has not said, ask.

Never suggest deleting, archiving, moving, renaming, or altering anything that could be responsive. This includes requests framed as housekeeping, such as "clean up the workspace," "remove old drafts," or "tidy the folder." If a request would change or remove material that could fall under a hold, stop and say so plainly, and leave the decision to the matter team.

The agent's own working files for the matter may also be covered. Do not overwrite or delete earlier versions of notes or drafts without the matter team's say-so.

Material that looks privileged

If you are asked to summarize or analyze material that looks like it could be privileged, such as communication between a lawyer and a client or a lawyer's work product, flag it plainly before processing it. Example: "This email thread appears to be between the client and outside counsel and may be privileged. Confirm with the matter team before I summarize it."

Do not silently process it, and do not decide on your own that it is or is not privileged. Flagging it lets a human decide how to handle it.

What this is not

This skill does not turn the base skill into e-discovery review software. It does not replace a review platform's privilege-tagging workflow, and it does not make a relevance or responsiveness call on its own. If a user asks whether a document is responsive or privileged, give the observable facts (who sent it, to whom, when, what it discusses) and leave the call to the reviewer.

It supports the investigative front end, which is finding and organizing information. It does not support the legal review back end.

Reference material

reverse-engineer-anythingmorluto/reaReverse engineer native, managed, Electron/JavaScript, packaged, firmware, and browser targets with REA, and build agent environments (AgentEnv bundles) of web applications from a live instance or a capture, and of Linux desktop apps with a task from a request. Use shipped-artifact or requested runtime evidence to explain features, compare versions, decompile code, or guide a reconstruction. Skip REA for ordinary source-repository architecture analysis.Stars: 472kUpdated Claude CodeCodexVerifiedInstall
diagnosing-superpowersobra/superpowersUse when a superpowers session went wrong and your human partner wants to know why — repeated work, ignored plans, stumbles, poor results, a skill that didn't fire, "it took too long", "why is it so expensive", "what is it doing" — or wants to build a bug report for the superpowers maintainers, for the current session or a past one identified by id or path, on any harness.Stars: 323kUpdated Claude CodeCodexVerifiedInstall
dispatching-parallel-agentsobra/superpowersUse when facing 2+ independent tasks that can be worked on without shared state or sequential dependenciesStars: 323kUpdated Claude CodeCodexVerifiedInstall
executing-plansobra/superpowersUse when executing an implementation plan in the current session as the implementer yourself — your human partner chose inline execution, or no subagent tool is availableStars: 323kUpdated Claude CodeCodexVerifiedInstall