القائمة
End-to-end eval-suite orchestration with the `eval` command: run -> grade -> compare -> calibrate, using strict JSON contracts and schema discovery.
التثبيت باستخدام Codex أو Claude انسخ هذا Prompt والصقه في Codex أو Claude أو مساعد آخر ليراجع صفحة Skill ويثبّتها لك.
Interview the user relentlessly about a plan or design until reaching shared understanding, resolving each branch of the decision tree. Use when user wants to stress-test a plan, get grilled on their design, or mentions "grill me".
Use test-driven development for behavior-changing feature or fix work, and whenever the user mentions TDD, test-first, red-green-refactor, tracer bullets, integration tests, or public-interface behavior tests. Skip for docs-only, path-only rename, formatting-only, or purely mechanical chores unless explicitly requested.
Analyze graded `trial_result` JSONL outputs from the eval pipeline, including reliability and custom slicing.
Build command-only adapters for `eval` run mode using strict stdin/stdout JSON contracts.
استنادا إلى تصنيف SOC المهني
| name | trial-runner |
| description | End-to-end eval-suite orchestration with the `eval` command: run -> grade -> compare -> calibrate, using strict JSON contracts and schema discovery. |
| license | ISC |
Use this skill to build or operate an eval suite around:
agent-eval-harness eval '<json>'
grill-me first to clarify:
tdd to build in vertical slices:
run smoke testgrade smoke testcompare checkcalibrate checkeval --schema taskeval --schema adapter-inputeval --schema adapter-outputeval --schema grader-inputeval --schema grader-outputeval --schema trial-rowagent-eval-harness eval '{"mode":"run",...}' > raw.jsonl
agent-eval-harness eval '{"mode":"grade",...}' < raw.jsonl > graded.jsonl
agent-eval-harness eval '{"mode":"compare",...}'
agent-eval-harness eval '{"mode":"calibrate",...}'
run and grade stream compact trial_result JSONL rows to stdout.
compare and calibrate emit one bounded JSON object.