giggleliu/Agentic Tests

Run AI-driven agentic tests on your project using OpenCode, Codex, or Claude Code

View on GitHub

Trust Signals

Scorecard Score
not yet scored
Maintenance Recency
Stale
License
None
namedescriptionrequireddefault
runnerAgent runner: "opencode", "codex", or "claude-code"noopencode
providerLLM provider — anthropic, openai, moonshot, etc.yes
modelModel to useno""
modeTest mode: "feature" (test project features), "skill" (test skill flows), or "both"nofeature
featuresComma-separated list of features/skills to testyes
profiles-dirDirectory containing saved agent profilesnodocs/agent-profiles
skillsSkills to install (comma-separated). Bare names resolve to built-in skills; paths resolve to local directories containing SKILL.mdnotest-skill,test-feature,create-profile
skill-reposComma-separated GitHub repos containing skills to clone and install (e.g., "owner/repo,owner/repo2"). The AI agent reads each repo README to install skills.no""
extra-promptExtra instructions appended to each test prompt (e.g., "focus on error handling", "test as a beginner")no""
issue-numberIssue or PR number to post results to (optional)no""
namedescription
features-testedComma-separated feature names tested
pass"true" if no critical issues found