Add greenproof to §8 (verifiers)

README.md

@@ -311,6 +311,7 @@ Most "awesome" lists are link dumps. This one is **annotated and verified**: eve
 - **[A Survey on LLM-as-a-Judge](https://arxiv.org/abs/2411.15594)** — Jiawei Gu et al. — <https://arxiv.org/abs/2411.15594> · *paper* — The most-cited survey organizing the LLM-judge space (bias taxonomy, reliability methods, agreement metrics). Serves as the one-stop map/bibliography the section currently lacks.
 - **[One Token to Fool LLM-as-a-Judge](https://arxiv.org/abs/2507.08794)** — Yulai Zhao, Haolin Liu, Dian Yu et al. (Tencent AI Lab / Princeton) — <https://arxiv.org/abs/2507.08794> · *paper* — Shows 'master-key' tokens (a colon, 'Solution:') trigger false-positive rewards up to 80% even on GPT-o1/Claude-4 judges, plus a robust Master-RM fix. Core evidence on judge/verifier reward-hacking fragility. 🆕
 - **[Weaver: Closing the Generation-Verification Gap with Weak Verifiers](https://hazyresearch.stanford.edu/blog/2025-06-18-weaver)** — Jon Saad-Falcon et al. — Stanford Hazy Research / Scaling Intelligence — <https://hazyresearch.stanford.edu/blog/2025-06-18-weaver> · *blog* — Directly operationalizes 'verifiable vs judgeable': aggregates many weak judges/reward models (unlabeled) to shrink the generator-verifier gap, reaching o3-mini accuracy from Llama-3.3-70B. Paper: arxiv.org/abs/2506.18203. 🆕
+- **[greenproof](https://github.com/zxyasfas/greenproof)** — <https://github.com/zxyasfas/greenproof> · *tool* — Deterministic (no LLM) verifier for whether a coding agent's green test run is real: snapshots the repo's tests before the agent runs, then reruns the agent's current code against the untouched originals — if the code fails them, the green depended on the agent's test edits, not a real fix. Self-reported: caught every constructed cheat and flagged no legitimate edit across 15 hand-built scenarios on 5 real Python repos; not a statistical benchmark. 🆕
 - **[Agent-as-a-Judge: Evaluate Agents with Agents](https://arxiv.org/abs/2410.10934)** — Mingchen Zhuge et al. (Meta AI / KAUST) — <https://arxiv.org/abs/2410.10934> · *paper* — Extends LLM-as-judge to agentic trajectories—grading intermediate steps, not just final outputs—with the DevAI benchmark. The agent-specific evaluation case this agent-evals library specifically needs.
 - **[VerifyBench: A Systematic Benchmark for Evaluating Reasoning Verifiers Across Domains](https://arxiv.org/abs/2507.09884)** — Various (AAAI 2026) — <https://arxiv.org/abs/2507.09884> · *benchmark* — Cross-domain benchmark exposing verifier precision/recall trade-offs (specialized verifiers high-accuracy but low-recall; general models inclusive but unstable). Quantifies how trustworthy a verifier actually is for RLVR. 🆕
 - **[Enhancing LLM-as-a-Judge with Grading Notes / From Pilot to Production with Custom Judges](https://www.databricks.com/blog/pilot-production-custom-judges)** — Databricks (Mosaic Research) — <https://www.databricks.com/blog/pilot-production-custom-judges> · *blog* — Enterprise-grade judge-building playbook: 20-30 calibration examples, batched SME annotation, Krippendorff's alpha agreement gating—a production-side complement to the Hamel/Shankar academic alignment loop. 🆕