karpathy-patterns

# Karpathy Patterns — Autonomous Agent Design Principles > **Source**: `karpathy/autoresearch` (Python, 3 files) + `karpathy/agenthub` (Go, single binary) > **Philosophy**: Quality is guaranteed by constraints, not infrastructure. > **Historical provenance**: Blog 22 (`22-karpathy-autoresearch-autonomous-ml-loop.md`) > and Blog 23 (`23-karpathy-agenthub-agent-native-infrastructure.md`) are absent > from this checkout; the former links are not maintained source references. ## Current GEODE applicability — reviewed 2026-09-06 The upstream pattern notes below are conceptual references. Their GEODE counterpart paragraphs are historical, not current implementation instructions: old node paths, iteration limits, layer numbers, and event counts must not be copied into new work. Ground present-day application in [AGENTS.md](../../../AGENTS.md), [code conventions](../../../docs/architecture/naming-conventions.md), and [runtime authority](../../../docs/architecture/coding-runtime-authority.md). `core` owns the runtime, `evals` measurement, and `evolve` scaffold search and promotion. Use the [architecture roadmap](../../../docs/architecture/extensibility-roadmap.md) for current GAP status; this skill is not another status ledger. The upstream Git/reset and unattended-run examples do not authorize discarding user work or starting live experiments. GEODE's workflow and frozen evaluation contract govern those actions. ## 10 Pattern Overview | # | Pattern | One-line Principle | Freedom | Source | |---|---------|-------------------|:-------:|--------| | P1 | Constraint-based design | Define "what cannot be done" first | Guardrail | autoresearch | | P2 | Single-file constraint | Modification surface = 1 file (or minimal unit) | Guardrail | autoresearch | | P3 | Fixed time budget | Compare fairly by wall clock, not steps | Guideline | autoresearch | | P4 | Ratchet mechanism | Keep only improvements, auto-revert on degradation | Guardrail | autoresearch | | P5 | Git as State Machine | Commit=experiment, reset=discard, tip=best solution | Guideline | autoresearch | | P6 | Context Budget management | Protect context via redirect + selective extraction | Guardrail | autoresearch | | P7 | program.md interface | Changing agent behavior = editing the instruction document | Inspiration | autoresearch | | P8 | Dumb Platform | Platform only stores, orchestration goes in prompts | Inspiration | AgentHub | | P9 | Branchless DAG | Agent collaboration via unnamed commit DAGs | Inspiration | AgentHub | | P10 | Simplicity Selection | Code deletion improvement > code addition improvement | Guideline | autoresearch | > **Freedom legend**: Guardrail = must follow for safety · Guideline = preferred but use judgment · Inspiration = conceptual reference --- ## P1. Constraint-based Design Restrict agent freedom to the necessary minimum. autoresearch constraints: 3 files, only train.py modifiable, 5-minute wall-clock, no package installation, single val_bpb metric. **Judgment**: When designing an agent, did you define "what cannot be done" before "what can be done"? **Historical GEODE counterpart — not current**: Node contracts (output keys restriction, `core/nodes/*.py`), Clean Context (analyses blocked, `analysts.py:417`), Confidence Gate ≥ 0.7 + max 5 iter (`graph.py:66-68`). --- ## P2. Single-file Constraint ``` autoresearch: train.py (~630 lines) = the only modification target → Entire code fits in context window, holistic understanding, diff=experiment record ``` **Judgment**: | Scenario | Applicable? | |----------|:-----------:| | Autonomous experiment / config optimization | O | | Large-scale refactoring / multi-module changes | X | **Historical GEODE counterpart — not current**: Each Analyst/Evaluator had independent prompts + independent output models. One node did not modify another node's prompts. --- ## P3. Fixed Time Budget ```python TRAINING_BUDGET_SECONDS = 300 # Efficient architecture = more steps (automatic reward) ``` Instead of "N iterations," use "do your best within T minutes" → the agent optimizes its own efficiency. **Historical GEODE counterpart — not current**: Then iteration-based (max 5); the note proposed node timeout + partial-result handling before wall-clock control. --- ## P4. Ratchet Mechanism ``` LOOP: modify → evaluate → if better: keep, else: revert ``` **Strength**: Safe for unattended overnight execution. **Weakness**: Can get stuck in local optima. **Mitigation**: Diversity Forcing (5 consecutive same type → forced switch), Simulated Annealing, Multi-branch (AgentHub DAG), Meta-optimization (program.md self-modification). **Historical GEODE counterpart — not current**: 5-Phase RLHF feedback loop (`automation/feedback_loop.py`). Broader exploration than ratchet (expert panel) + weaker convergence guarantee. --- ## P5. Git as State Machine ``` Commit = experiment record Branch tip = best solution git reset = discard failure ``` Zero infrastructure cost. **Weakness**: `git reset` loses failure records → risk of repeating the same failures. **Historical GEODE counterpart — not current**: 3-Tier Memory (`memory/organization.py`, `project.py`, `session.py`) was presented as retaining failure history via hierarchical TTL. --- ## P6. Context Budget Management ```bash uv run train.py > run.log 2>&1 # L1: Block (0 context consumption) grep "^val_bpb:" run.log # L2: Extract (only 2 lines) # L3: Summarize → 1-bit judgment (improved/degraded) ``` **Historical GEODE counterpart — not current**: Clean Context — existing analyses excluded from Send API (`analysts.py:418-434`). Session TTL (`session.py:43-51`). PromptAssembler assembled selected information per node (`prompt_assembler.py:48-110`). --- ## P7. program.md Interface program.md = agent instruction document. Composed of Setup (initialization) + Experimentation (loop protocol) + Constraints (prohibitions) + Preferences (direction) + Style (quality standards). **Key point**: The quality of program.md determines the quality of the agent's research. To change behavior, modify the instruction document, not the code. **Historical GEODE counterpart — not current inventory**: CLAUDE.md (project instruction document) + skill system (domain-specific expert instruction documents) + HookSystem 26 events (`hooks.py:19-62`). --- ## P8. Dumb Platform ``` Smart Platform (GEODE/OpenClaw): Platform = routing + concurrency + events + orchestration Dumb Platform (AgentHub): Platform = storage + delivery only, orchestration in prompts ``` **Judgment**: | Scenario | Recommendation | |----------|---------------| | Deterministic ordering / SLA | Smart | | Frequent orchestration changes / open-ended exploration | Dumb | | **Hybrid** | Pipeline is Smart, inter-agent discussion is Dumb | **Historical GEODE counterpart — not current**: Classified as Smart Platform; the note proposed partial Dumb elements with a future L6 Custom Agent layer. That layer map is not the current architecture. --- ## P9. Branchless DAG A DAG where commits branch out in all directions, without branches/PRs/merges. Core operations: `leaves` (frontier), `lineage` (ancestor path), `children` (direct descendants). **Historical GEODE counterpart — not current**: TaskSystem's `get_ready_tasks()` (`task_system.py:116-120`) was compared with `leaves` — pending tasks with fulfilled dependencies as frontier nodes. --- ## P10. Simplicity Selection ``` program.md: "Add 20 lines for 0.001 improvement? Reject. Delete code for 0.001 improvement? Always accept." ``` | Change | Improvement | Verdict | |--------|-------------|---------| | Code deletion | Marginal | **Accept** | | Clean addition | Meaningful | Accept | | Hacky addition | Marginal | **Reject** | LLMs are inherently biased toward adding code. The instruction document must explicitly include "prefer simple solutions." **Historical GEODE counterpart — not current prompt text**: Compared with the system prompt's "Avoid over-engineering. Only make changes that are directly requested" principle. --- ## Pattern Relationships ``` P1 Constraint-based design ─── Top-level principle ├── P2 Single file (code level) ├── P3 Fixed time (resource level) └── P10 Simplicity (quality criteria) P4 Ratchet ─── Safe autonomous execution └── P5 Git State Machine (implementation mechanism) P6 Context Budget ─── Sustained long-running execution └── P7 program.md (human-agent interface) P8 Dumb Platform ─── Multi-agent scaling └── P9 Branchless DAG (implementation pattern) ``` ## Scale-based Application Guide | Scale | Applicable Patterns | Not Applicable | |-------|-------------------|----------------| | Single agent, single task | P1, P2, P4, P5 | P8, P9 | | Single agent, overnight autonomous | P1-P7, P10 | P8, P9 | | Multi-agent, exploratory | P1, P4, P6, P8, P9 | P2 | | Multi-agent, production | P1, P3, P4, P6 + Smart | P8 | ## Anti-patterns | Anti-pattern | Violates | Symptom | |-------------|----------|---------| | "All files modifiable" | P2 | Fragmented changes, broken dependencies | | "Unlimited execution" | P3 | Cost explosion, meaningless exploration | | "All results into context" | P6 | Context exhaustion, early termination | | "Platform controls everything" | P8 | Lost flexibility, deployment bottleneck | | "Accept any improvement" | P10 | Complexity accumulation | | "Failure records not preserved" | P5 | Repeating the same failures |