/trip — A Synthesis Pass Over Your Second Brain

A structured, phased methodology for using an AI assistant to surface connections in a personal knowledge base that wouldn't surface from default search.

The skill borrows the cognitive arc of a classic high-dose psychedelic experience — constraint loosening → associative widening → peak coherence → reconsolidation → durable change — as scaffolding for AI-assisted synthesis. The AI isn't tripping. The skill uses the shape of a trip as a cognitive scaffold because the shape, documented across sixty years of clinical research, happens to be a useful structure for producing connections that don't surface from default retrieval.

What's in this package

• SKILL.md — the runnable skill file (v1.2.1). Drop it into your AI assistant's skills directory (e.g. .claude/skills/trip/) and invoke with /trip [your question]. Seven phases. Hard rules. Phase 0 is a conversational seven-question intake modeled on MAPS-protocol psychedelic-therapy practice — the agent interviews you into the trip instead of taking parameters upfront. ~2 hours of work + 30 days of integration.
• EXPLAINER.md — the plain-language version. No jargon. Lift-able as a tweet thread, Substack callout, or one-pager. Use this to explain the skill to someone who's never heard of it.
• METHODOLOGY.md — the longer methodology document. Reads as a written-prose essay rather than a runnable spec. Includes the lineage, the research foundations, and the design rationale.
• PRIOR-ART.md — the literature assessment conducted before public release. What exists, what's adjacent but different, what appears genuinely novel. Verdict, recommended citations, differentiation language, and the caveats that need real verification before publishing strong novelty claims.

Prerequisites

You need three things to run /trip:

1. A personal knowledge base — a folder of markdown files. Your notes, your saved articles, your journal entries, anything you've actually read. The skill is a procedure run on top of a knowledge base; it doesn't create one.
   • If you don't have a knowledge base yet, the fastest path is the second-brain skill pack by Nicholas Spisak — an Obsidian-based implementation of Andrej Karpathy's LLM Wiki pattern. One command to install, about ten minutes to set up.
2. An AI assistant that supports skill files. Claude Code is the canonical environment; Cursor, Codex CLI, Gemini CLI, and 40+ other agent harnesses also work.
3. A specific question. Vague intentions produce vague trips. The skill's hard rule.

Installation

# Option A — Claude Code
mkdir -p .claude/skills/trip
cp SKILL.md .claude/skills/trip/SKILL.md

# Option B — other agent harnesses
# Place SKILL.md in your agent's skills directory.
# Naming conventions vary; check your harness's documentation.

If your agent doesn't support skill files, you can still run /trip manually — see METHODOLOGY.md for the phase-by-phase prompts you can paste in sequence.

First run

/trip

The agent runs the v1.2 progressive intake — seven questions, one at a time. Your first time through, expect ~20–30 minutes for Phase 0 (the intake), then another ~90 minutes for Phases 1–5. Phase 6 runs over the following 30 days as three brief check-ins.

A first-run tip: pick a question your notes seem to hover around but never quite crack. Don't bring something already resolved.

How it works (in 30 seconds)

Seven phases:

1. Question — set your intention.
2. Go wide — pull notes from unrelated areas, look for weird connections.
3. Go deep — re-read slowly, surface buried material, hold strange juxtapositions.
4. Make a big claim — force a unification under suspended critical filtering.
5. Check yourself — reverse the stance, verify, cut.
6. Write it down — produce a synthesis artifact.
7. Come back to it — at +24h, +7d, +30d. Track what consolidates.

The skill is opinionated about Phase 6. Most AI synthesis tools skip integration. Without the 30-day window, the work doesn't stick.

Three findings from research that shape the design

1. Set and setting predict outcomes more than dose. Intention matters more than searching effort.
2. Peak intensity predicts outcome. Going deep beats going wide.
3. Integration is where change consolidates. The artifact alone doesn't change you. Using it does.

Sources: Carhart-Harris & Nutt (2017); Roseman, Nutt & Carhart-Harris (2018); Griffiths group (multiple); MAPS protocols. All real, all citable.

Who this is for

Use the skill if you:

• Keep notes / bookmarks / a journal you actually read back.
• Have an AI assistant you use against those notes (Claude Code, ChatGPT, NotebookLM, whatever — the methodology is portable).
• Have a specific question you've been chewing on but haven't cracked.
• Can commit to 30 days of follow-up. (This is the part most people skip; it's also the part that makes the work stick.)

Don't use the skill if you:

• Need a fast answer.
• Don't have a specific intention.
• Are looking for novelty without follow-through.

Lineage

This work stands on a thick lineage:

• The tools-for-thought tradition — Bush ("As We May Think," 1945); Engelbart ("Augmenting Human Intellect," 1962); Nelson (Xanadu); Luhmann's Zettelkasten codified by Ahrens (How to Take Smart Notes, 2017); Forte (Building a Second Brain, 2022); Matuschak (Evergreen Notes; Quantum Country with Nielsen).
• The LLM-as-knowledge-substrate frame — Andrej Karpathy's LLM Wiki gist (2024–2025). The single most important contemporary citation. The substrate the /trip skill assumes.
• Phased-cognition ancestry — Graham Wallas, The Art of Thought (1926). The four-stage model of creative thinking (preparation → incubation → illumination → verification). The trip skill's arc is structurally homologous to Wallas's model; any phased synthesis methodology has to acknowledge him as the ancestor.
• Sixty years of clinical psychedelic research — Hofmann; Grof; Pahnke; Griffiths; Carhart-Harris; MAPS; the MDMA-couples-therapy lineage (Leo Zeff, Greer & Tolbert 1986, the Shulgins).
• The "borrow shape, drop modality" precedent — Stan Grof's Holotropic Breathwork (formalized ~1976). Grof literally did the move /trip is doing — one step less far. He took the LSD phase arc and built a non-drug method that produces similar outcomes via hyperventilation, music, and bodywork. The /trip skill extends Grof's move one step further: borrow the phase arc, drop the chemistry and the experience. The agent doesn't trip; only the prompt sequence has phases.
• Methodology-as-method-call framing — Garry Tan's Skill Files framework (2026).

What's new: The /trip skill's contribution is the combination — Karpathy's substrate + Wallas-style phasing operationalized as a prompt protocol + the specific arc from clinical psychedelic research + Grof's "borrow shape" precedent extended one step further + MAPS-style integration windows applied to synthesis frames (rather than to facts, as spaced repetition does). To the curator's training-knowledge survey, no equivalent staged methodology has been formalized for AI-assisted synthesis on personal knowledge bases. See PRIOR-ART.md for the full assessment, the recommended citations, and the caveats that need live-web verification before publishing strong novelty claims.

License

MIT. Use it, modify it, fork it. Attribution appreciated. The methodology is mine; the credit belongs to the clinical research community that mapped the cognitive arc.

About

Built by Ryan Batiste, 2026-05-14. Written about at ryanbatiste.substack.com. The companion essay — Tripping Through My Second Brain — documents the first run of the skill on itself (recursive). The skill is the receipt for the essay's claim.

Questions, runs, or improvements: drop a note. Issues, forks, and pull requests welcome at whatever hosting this lands on.