CODE Framework
Definition
The CODE framework is a four-step personal knowledge management processing loop created by Tiago Forte: Capture, Organize, Distill, and Express. It is the core methodology of the Building a Second Brain system, designed to turn consumed information into creative output and concrete results.
Key Points
- Capture — save the most important information (ideas, insights, resources) that you encounter. Rules of thumb: think like a curator, keep only what resonates, use capture tools.
- Organize — file notes by actionability, not subject hierarchy. Notes are placed in PARA categories (Projects, Areas, Resources, Archives) based on how soon they will be used.
- Distill — extract the essence from accumulated notes via progressive summarization (multiple layers of condensation), opportunistic distilling (add value every time you touch a note), and designing notes for your future self.
- Express — use the organized knowledge to produce tangible output (writing, products, projects). Create smaller reusable “Intermediate Packets” and share work with the world.
AI-Native Adaptation (Forte 2024)
In Forte’s 2024 AI analysis, AI “radically disrupts the middle two steps” — Organize and Distill — because AI can extract structure from unstructured text and summarize large volumes rapidly. What AI does not change is Capture (human decides what matters) and Express (human adds voice, judgment, finishing direction).
The LLM Wiki pattern operationalizes exactly these AI-disrupted steps: the human does Capture (drops raw material into raw/), and the LLM handles Organize + Distill during ingest. ^[raw/external/angelo-lima-fr-karpathy-second-brain-obsidian-claude-en-d7284e8e.md]
Mapping to LLM-Wiki operations
| CODE Step | LLM-Wiki / OKF counterpart | Who does it |
|---|---|---|
| Capture | raw/ folder | Human |
| Organize | wiki/ folder structure (concepts, entities, etc.) | LLM (with human oversight) |
| Distill | LLM synthesis, cross-linking, contradiction detection | LLM |
| Express | Querying, briefings, using compiled knowledge | Human + LLM |
The CLAUDE.md / AGENTS.md schema file acts as the “contract” that replaces Forte’s manual organizational rules with agent-enforceable behavior. ^[raw/external/github-com-second-brain-for-claude-4c9d7132.md]
Related Concepts
- para-method — the organizing system that pairs with CODE (Projects, Areas, Resources, Archive)
- second-brain — the system CODE is designed to build and maintain
- llm-wiki — the LLM Wiki pattern that automates the Organize and Distill steps
- ingest — the LLM Wiki operation that performs CODE’s Organize + Distill automatically
Implications
The convergence is structural, not replacement. Forte’s 2024 analysis already anticipated that AI would take over Organize and Distill — the LLM Wiki pattern is the first reproducible implementation of that vision. Practitioners who combine CODE+PARA with LLM Wiki do not abandon Forte’s framework; they delegate its two most maintenance-intensive steps to the agent while retaining human judgment at Capture and Express. The governance file (CLAUDE.md/AGENTS.md) replaces Forte’s manual “rules of the game” with an enforceable contract, but the epistemic division of labor remains the same.
Open Questions
- Does fully automated Organization+Distillation risk losing the personal meaning-making that Forte emphasizes as central to Express?
- How does the quality of LLM distillation compare to human progressive summarization over long time horizons?
Sources
^[raw/external/fortelabs-com-basboverview-ee1d6e3d.md] ^[raw/external/angelo-lima-fr-karpathy-second-brain-obsidian-claude-en-d7284e8e.md] ^[raw/external/github-com-second-brain-for-claude-4c9d7132.md]