[uipreliga]

Adds Coder Eval to Benchmark, alphabetically between ASQI Engineer and LiveBench.

Disclosure: I maintain this project. Happy to reword or move it — Tool would also be reasonable if you read it as tooling rather than a benchmark.

Why it belongs next to these

The section already carries lm-evaluation-harness, OpenCompass, opik, terminal-bench and skill-optimizer. Coder Eval is closest to that last pair: rather than scoring a model on a fixed dataset, it holds the task suite fixed and varies the harness and the skills, then reports how each configuration did.

Its specific measurement is skill activation — whether an agent's skill actually fires on the prompts it claims to handle, and stays quiet on the rest — expressed as suite-level precision / recall / F1 over a labelled dataset, with thresholds that can fail a CI run when they regress.

• Repo: https://github.com/UiPath/coder_eval — Apache-2.0, 118★, Python
• Docs: https://coder-eval.com/docs
• Backends: Claude Code, Codex, OpenCode, Antigravity; Anthropic / Bedrock / LiteLLM routing

I matched the section's entry format exactly (bold name, colon, description, then the stars / contributors / last-commit badges) and kept alphabetical order. One line, no other changes.