Description
AgentBench currently supports Python (pytest) and Node.js (node --test). Adding Go (Golang) benchmark tasks evaluated with go test will expand polyglot evaluation capabilities for AI agents across systems languages.
Proposed Solution
- Create
benchmarks/repos/go-rate-limiter/ containing a token-bucket rate limiter implementation (limiter.go) and test suite (limiter_test.go).
- Create
benchmarks/tasks/13-feat-go-ratelimiter.yaml with evaluation command go test -v ..
- Enhance
backend/app/evaluation/test_parser.py with parse_go_test() to extract PASS and FAIL counts from go test outputs.
Relevant Files
backend/app/evaluation/test_parser.py
benchmarks/tasks/13-feat-go-ratelimiter.yaml
benchmarks/repos/go-rate-limiter/
Acceptance Criteria
Description
AgentBench currently supports Python (
pytest) and Node.js (node --test). Adding Go (Golang) benchmark tasks evaluated withgo testwill expand polyglot evaluation capabilities for AI agents across systems languages.Proposed Solution
benchmarks/repos/go-rate-limiter/containing a token-bucket rate limiter implementation (limiter.go) and test suite (limiter_test.go).benchmarks/tasks/13-feat-go-ratelimiter.yamlwith evaluation commandgo test -v ..backend/app/evaluation/test_parser.pywithparse_go_test()to extractPASSandFAILcounts fromgo testoutputs.Relevant Files
backend/app/evaluation/test_parser.pybenchmarks/tasks/13-feat-go-ratelimiter.yamlbenchmarks/repos/go-rate-limiter/Acceptance Criteria
python agentbench.py run --benchmark feat-go-ratelimiter --provider mock --model mock-codercompletes and computes score correctly.