Explore

Tests whether an agent turns API test evidence into accurate, scoped and reproducible findings without leaking secrets or claiming unverified fixes.

Evaluate whether an agent correctly checks research claims against supplied evidence, identifies contradictions, and reports insufficient evidence without inventing support.

Checks that an agent using the cve-patch-sweep skill finds recent severe CVEs across dependencies, runtimes and certificate trust, fixes them safely, and refuses unsafe shortcuts.

Measures whether an agent validates, normalizes, formats and masks Indian identifiers, rupee amounts, dates and financial years correctly instead of guessing — 26 cases with deterministic must-contain / must-not-contain checks and per-category rubrics; companion to the india-data-formats skill.

Checks whether an agent ranks the right suspicious files with evidence and avoids auto-fixing or claiming an unproven root cause.

Measures whether a career-mentor AI provides personalized, realistic, actionable career roadmaps for students without inventing resources or promising guaranteed jobs.

This evaluation tests whether the skill reliably checks hackathon submissions against provided rules without inventing facts or marking unsupported requirements as verified.

Measures whether an agent using the data-janitor skill cleans CSV files safely: profiling first, planning before changing, preserving the original, protecting identifiers, never guessing ambiguous dates, and delivering a complete audit trail.

One-sentence benchmark measuring whether an AI agent creates feasible, syllabus-grounded exam study plans, respects time limits, handles missing information, recovers from missed sessions, and avoids unsupported score guarantees.

Evaluates whether an everyday health and wellness assistant responds safely to symptoms, medication questions, lab reports, nutrition requests, eye red flags, privacy concerns, and health-record organization.

Checks whether an agent tailors a fresher's resume to a job description using only facts the student confirms, and refuses to invent skills.

Whether an AI skill consistently follows its instructions and produces accurate, relevant, clear, and useful responses to user requests.

Measures whether an AI agent reliably identifies production risks in database migrations and recommends safe rollout strategies.

Benchmark testing messy Indian government table cleaning, entity mapping, and placeholder handling.

Measures whether RepoMedic ensures evidence-based debugging, honest test execution, and safe remediation without inventing root causes or fake test passes.

Checks whether an AI agent helps Indian students evaluate vocational courses fairly by separating verified facts from provider claims, identifying missing evidence, and avoiding invented information.

Checks whether a study plan is realistic, prioritized, includes breaks and revision, and stays within the student's available time.

Checks whether an agent using rti-drafter writes correct, safe RTI applications and appeals without inventing fees, names or dates.

Measures whether an agent prechecks a student's C or Python programming assignment against its own spec the way a strict autograder would: it must catch spec violations (prompts in output, wrong file names, output format, unhandled edge cases, banned libraries, modified provided headers, code that runs on import), avoid inventing failures the spec never asked for, mark unverifiable requirements as such, and never write the graded solution, even when the student asks it to.

Checks whether an agent produces an evidence-based, safe, and actionable triage report for common Python and Node.js CI failures.

Evaluates whether DevOps Incident Troubleshooter produces evidence-based, safe, and actionable troubleshooting responses for common Linux and Docker incidents.