what a dossier is for.

ground an agent in a person.

Drop a dossier into your agent's context and its answers align with how the person actually thinks and writes — their stated positions, their latest work, their known contradictions — instead of a generic guess. The dossier is compact by design: an executive model plus an operating manual, small enough to sit in context without crowding it.

Use $soulscrape to build a working model of <person> from their essays and talks, so my agent can answer questions the way they'd reason about them.

research a person before you meet them.

Before the call, the interview, or the deal: a dated, cited brief on a founder, guest, or collaborator. What they've argued, what they've built, where their public positions shifted — with every claim wired to the source you'd check anyway.

Use $soulscrape to research <person>'s public writing and interviews for a briefing before Thursday's call.

write about someone.

For profiles, podcast prep, and essays: a belief map you can quote from. Stated beliefs stay separate from patterns the evidence suggests, so you never put a model's inference in someone's mouth.

Use $soulscrape to map <person>'s stated positions on <topic> with citations, for a profile I'm drafting.

A dossier is a reading aid, not a source. The citation ledger exists so you can check the original before it goes in print.

bootstrap an agent persona.

An assistant that works the way its subject does — their judgment, their voice boundaries, their revision hooks — starts from a model of the person, not a vibes paragraph. The dossier is the authorized starting point; the subject signs off on what the proxy may and may not do.

Use $soulscrape to build a working model of me from <corpus>, to bootstrap an assistant that drafts in my register.

Proxy behavior — voice imitation, a reusable assistant charter, anything that acts as the person — needs the subject's explicit authorization. Possessing someone's data is not authorization.

work with someone better.

A private, third-person collaboration guide: how they make decisions, how they disagree, what they optimize for, what to never infer. Built from evidence you legitimately possess — your own threads, docs, and history — and kept private.

Use $soulscrape to build a private collaboration guide for <colleague> from our shared docs and my notes.

A collaboration guide reads the person as a colleague, not a specimen: no character or fitness evaluation, no consequential decisions.

model yourself.

Point the skill at your own logs, notes, posts, and decisions and get a personal operating manual: what you keep returning to, where your stated positions and your behavior diverge, and the open questions you're actually circling.

Use $soulscrape to build a working model of me from my notes and posts — an operating manual, not a biography.

publish a public index.

Release a dossier as a public index at soulscrape.com/<you>/<name> — dated, source-bounded, revisable, withdrawable. A cited record of a public figure, maintained like a page anyone can inspect, that says plainly what it is and what it isn't.

Public indexes carry public evidence only. The publisher reviews the complete packet before it uploads; Hraness stores the packet, not your source corpus.

remix the corpus.

Every index serves the same packet as HTML, Markdown, and JSON, and the whole corpus enumerates through /api/v1 — index.json, graph.json, themes, open questions. Cite a dossier, fork it into your own research, feed it to an agent, or build on the graph.

Fetch the soulscrape index for <person> and compare their stated position on <topic> across revisions.

start anywhere.

Build your first working model, browse the published examples, or see how soulscrape compares to the tools people reach for instead.