harb/STATE.md
openhands cb6e6708b6 fix: \llm\-origin entries in manifest have null fitness and no evaluation path (#724)
- Add evaluate-seeds.sh: standalone script that reads manifest.jsonl,
  finds every entry with fitness: null, runs fitness.sh against each
  seed file, and atomically writes results back to manifest.jsonl.
  Supports --dry-run to preview without evaluating.
- Add comment to --diverse-seeds sampling in evolve.sh documenting that
  null-fitness seeds are included with effective_fitness=0 and that
  evaluate-seeds.sh should be run to score them.

Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
2026-03-15 03:08:29 +00:00

3 KiB

STATE.md — What harb currently is and does

  • [2026-03-13] Evolution pipeline works end-to-end: Push3 → transpile → compile → revm fitness evaluation → selection (#665)
  • [2026-03-13] Diverse seed generation for evolution population (#656)
  • [2026-03-13] Crossover operator for Push3 programs (#657)
  • [2026-03-13] Elitism preserves top N candidates unchanged across generations (#643)
  • [2026-03-13] Gas limit as evolutionary fitness pressure (#645)
  • [2026-03-13] Default bear outputs for crashed/broken Push3 programs (#651)
  • [2026-03-13] Normalized inputs for Push3 optimizer (0..1e18 indicators) (#649)
  • [2026-03-13] Bootstrap VWAP with seed trade during deployment (#633)
  • [2026-03-13] e2e tests skip for tools-only and docs-only PRs (#641)
  • [2026-03-13] Issue templates for bug, feature, push3-seed, refactor (#678)
  • [2026-03-13] revm fitness evaluator with UUPS bypass and graceful attack ops (#629)
  • [2026-03-12] Dark factory: dev-agent, review-agent, supervisor with cron */10 staggered
  • [2026-03-12] CI: single build-and-test pipeline + e2e with path filtering
  • [2026-03-12] Ponder indexing: transfers, mints, burns, staking, protocol stats
  • [2026-03-12] Landing page with LiveStats, WalletCard, contract addresses
  • [2026-03-12] Staking app with position dashboard and P&L tracking
  • [2026-03-12] OptimizerV3 with Push3 transpiler output injection
  • [2026-03-12] Three-position strategy: Floor, Anchor, Discovery
  • [2026-03-12] VWAPTracker for price oracle
  • [2026-03-12] Harberger tax staking mechanism
  • [2026-03-13] LLM seed — Momentum Follower optimizer (#695)
  • [2026-03-14] evolve.sh auto-incrementing per-run results directory (#752)
  • [2026-03-14] EVAL_MODE now defaults to revm (#751)
  • [2026-03-14] LLM seed — Defensive Floor Hugger optimizer (#672)
  • [2026-03-14] evolve.sh stale tmpdirs break subsequent runs (#750)
  • [2026-03-14] evolve.sh silences all batch-eval errors with 2>/dev/null (#749)
  • [2026-03-14] evolution-daemon.sh — perpetual evolution loop on DO box (#748)
  • [2026-03-14] No mainnet VWAP bootstrap runbook (#728)
  • [2026-03-14] fitness.sh individual-scoring path still silences errors (#766)
  • [2026-03-14] batch-eval.sh MANIFEST_DIR (mktemp -d) has no cleanup trap (#763)
  • [2026-03-14] bootstrap.sh anvil_setCode guard now targets correct feeDest 0xf6a3... (#760)
  • [2026-03-14] llm_contrarian.push3 AW=150/250 clamped to 100 — three rounds unaddressed (#756)
  • [2026-03-14] bootstrap.sh hardcodes BASE_SEPOLIA_LOCAL_FORK even on mainnet forks (#746)
  • [2026-03-14] remove MAX_ANCHOR_WIDTH clamp in ThreePositionStrategy (#783)
  • [2026-03-14] increase CALCULATE_PARAMS_GAS_LIMIT from 200k to 500k (#782)
  • [2026-03-15] add evolution run 8 champion to seed pool (#781)
  • [2026-03-15] fix FitnessEvaluator.t.sol broken on Base mainnet fork (#780)
  • [2026-03-15] No generic flag dispatch: only token_value_inflation is ever zero-rated (#723)
  • [2026-03-15] llm-origin entries in manifest have null fitness and no evaluation path (#724): evaluate-seeds.sh scores null-fitness seeds and writes results back to manifest.jsonl