/portrait — a deep dossier on a person

🧒 When reporting to a non-technical operator: end with a child-simple "In plain words" recap in their language.

An orchestrator: RECALL (everything we already have) → WEB (all social profiles) → a DR prompt (Alpha Protocol) → a card in the vault → synthesis once the operator brings the DR back. It chains /find + /ask + the Alpha Protocol + ingest, aimed at ONE person — a role model, a lead or a partner.

Step 0 — RECALL first (cheap, before the web)

Pull up EVERYTHING we already have (the recall-before-activity rule):

• Memory: grep through ~/.claude/.../memory/ for the name and its variants.
• Exact name: /find <name> → PYTHONUTF8=1 python "$IMPORTS_ROOT/namesearch/find_name.py" <name> (catches transliteration, wrong layout, typos).
• Meaning: /ask "<name> / topic" (RAG over the vault).
• A direct vault grep for the Latin slug (07-People\person-<slug>.md, 01-Conversations\...\<slug>, CRM leads, Facebook posts about them).
• Check for an existing card 07-People\person-<slug>.md (plus -2 duplicates) — ENRICH it, never breed a new one; duplicates → supersede (skill dedup).

Step 1 — WEB: all social profiles (in parallel)

WebSearch/WebFetch (Chrome MCP when paywalled). Build a table: X/Twitter, Telegram (channel + personal), Medium/Substack/blog, LinkedIn, Facebook, Instagram, YouTube, GitHub/Gist, Keybase, personal site, fund/company, podcasts, academic trail. Mark confidence and any dead/suspended links. Cross-verify identity (handle ↔ Keybase/GitHub proof) and separate namesakes by handle, topics and companies.

Step 2 — the DR prompt (Alpha Protocol; the deep DR is run by the operator in an external tool)

Emit a filled-in DEEP RESEARCH PROMPT as one clean block (subject, sources, CONTEXT from recall, 6-8 questions: psyche/character, mental models, evolution, influences, contradictions/blind spots, personal philosophy, "the imprint worth emulating"). Template: 08-Templates\deep-research-prompt-template.md. I do not run the deep DR myself — a light web pass is fine.

Step 3 — the card into the vault (backup → write → reindex)

1. python $IMPORTS_ROOT/vault_backup.py "<label>" BEFORE writing ([[vault-backup-rule]]).
2. 07-People\person-<latin-slug>.md (Latin slug! the native-script name goes into title:/aliases:): who they are · ⭐ why they matter to the owner + the link to their goals · a table of every social profile · a map of how they think · what the vault already holds (the recall index) · personal history (DMs/calls/CRM) · connections and concepts. At least 1 inbound link (no-orphan).
3. Reindex: python $IMPORTS_ROOT/brain_embed_update.py [--cpu] (15-minute cooldown → --force).

Step 4 — synthesis (once the operator brings the DR back)

Merge recall + DR → a Decision Memo 03-Insights\insight-* ("what to adopt / how they think") and weave it into the dossier. Original reports → _originals\<slug>-deep-research\ ([[preserve-originals-rule]]).

Gates

• We do not touch their social accounts from the outside; outreach to the person is draft-first + an explicit "yes" from the operator (Tier-2).
• Grunt work (collecting profiles, frequency analysis of a corpus) can run on cheap-model subagents; the synthesis and the map of thinking need the top model.
• Reference precedent: an earlier dossier built exactly this way (see the matching person-* and insight-* notes).

<!--kit-footer-->

---

Like this skill? It is part of second-brain-starter-kit: the second brain we built for ourselves and run every day at Palo Alto AI Research Lab. Install the whole set with npx skills add tonydzi/second-brain-starter-kit. Everything is open source and free, so take what you need.

Flagships worth a look on their own: secondop-panel (a second opinion from a panel of external models), claude-memory-tidy (stop your agent's memory from rotting), telegram-mcp-kit (your own Telegram over MCP in about 15 minutes).

Author: Anton Dziatkovskii, Palo Alto AI Research Lab. Telegram @tonydzi - WhatsApp +1 341 222 9178 - X @Tony_Stef_

Engineers: want to test-drive this setup? Message me. I hand out free starter seeds to engineers who test and report back, and custom skill requests are welcome.