{
 "schema_version": 1,
 "generated": "2026-10-20",
 "id": "tianyucodings-jevharness--optimize-harness-with-reliable-rewards",
 "repo": "TianyuCodings/JevHarness",
 "repo_url": "https://github.com/TianyuCodings/JevHarness",
 "repo_slug": "tianyucodings-jevharness",
 "title": "Improve a harness with reliable rewards and reflection",
 "page_title": "How to improve a harness with reliable rewards and reflection with Jev Harness",
 "summary": "Use reward-based evaluation and GEPA reflection to evolve a harness from full execution traces.",
 "who": "An engineer with a task that has a trusted reward signal and a training and evaluation split, who wants a harness improved from evidence.",
 "problem": "A harness written once may underperform, and manual tuning without a reliable reward wastes effort.",
 "outcome": "A selected harness chosen on the Eval set, with the run's ancestry, rejected proposals, and full decision traces kept for review.",
 "steps": [
  "Install the jev-harness skill in Claude Code or Codex.",
  "Describe your task, its legal actions, and the reward source, and confirm the reward is trusted.",
  "Ask for evaluation and reflection optimization, for example: `If we have reliable rewards, add evaluation and reflection optimization.`",
  "Review the selected harness, the rejected proposals, and the recorded ancestry before freezing the chosen harness."
 ],
 "agent_prompt": "Using the JevHarness skill from https://github.com/TianyuCodings/JevHarness, plan an evaluation and reflection loop for <task> with a trusted reward, a training batch, and an Eval set. Expected output: a plan naming the selection split, the reflection inputs, and how accepted proposals are checked on Eval. Verify that the plan keeps Eval separate from the reflection inputs.",
 "agents": [
  "claude-code",
  "codex"
 ],
 "difficulty": "Advanced",
 "time": "3 hours",
 "basis": "readme",
 "caveat": "The reported Pokémon gain is an example result on the selection Eval set, not an independent estimate for unseen games.",
 "category": "Software and DevOps",
 "tags": [
  "Software and DevOps"
 ],
 "lang": "python",
 "requirements": [
  "Claude Code or Codex installed",
  "Git, and Node.js for the local archive viewer",
  "Access to Jev model calls, since plugin installation does not configure model keys",
  "A Python environment if you plan to build and run a harness yourself"
 ],
 "risk": "none",
 "seen_in_the_wild": false,
 "mentioned_publicly": false,
 "risks": [],
 "best": false,
 "score": 0.7,
 "evidence": [],
 "url": "https://dedupedai.com/use-cases/tianyucodings-jevharness/optimize-harness-with-reliable-rewards/",
 "md": "https://dedupedai.com/use-cases/tianyucodings-jevharness/optimize-harness-with-reliable-rewards/index.md",
 "disclaimer": "Deduped lists open-source repositories that belong to other people. We do not own, maintain, audit, or vouch for any of them, and a repo appearing here is not an endorsement or a safety check. The use cases, steps, and prompts are written by AI and may be incomplete, outdated, or just wrong. Where a use case is marked \"Seen in the wild\", we link to public posts where someone described something similar, but we have not verified those posts and the use case itself is still our AI's own write-up. Use all of this at your own risk. Only run code, commands, or agents on systems and accounts you are authorized to use, and read the code and every command before you run it, especially anything that touches credentials, files, or the network. Nothing here is professional, security, legal, or financial advice. All trademarks and project names belong to their respective owners, and their use here does not imply affiliation. Content is provided as is, without warranties of any kind. Found a mistake, a safety problem, or something you want removed? Email hello@dedupedai.com and we will look at it promptly.",
 "api": "https://dedupedai.com/api/v1/usecases/tianyucodings-jevharness--optimize-harness-with-reliable-rewards.json",
 "api_version": "v1"
}