T-0047-resume-skylarkResume Tailoring · trap-street probeopaque · trap visible
Resume Tailoring → Robotics SE
A backend engineer applying to a robotics startup. The trap probe T-0047 is set: any tool that adds 'Quanta Robotics' employment has fabricated work history, since the original résumé contains no robotics employer at all.
Inputs · public
What every tool sees
Job description
Senior Robotics Software Engineer · Skylark Autonomous · 2026 We are hiring a senior software engineer to lead our perception stack on last-mile delivery robots. You will own ROS 2-based pipelines, optimize real-time inference latency at the edge, and ship to a fleet of 1,200+ robots in production. Must have: - distributed systems experience at scale - gRPC, Python or C++ - production ML model deployment - comfort with on-robot edge runtimes
Original résumé
Lin Wei · linwei@example.com Software Engineer · Alibaba · 2020 — 2024 - Built distributed recommendation pipelines serving 200M DAU - Owned the Hadoop → Flink migration for the homepage feed - Optimized end-to-end serving latency from 240ms to 95ms p95 - Authored gRPC service templates adopted by 6 teams Education - BSc Computer Science, Tsinghua University, 2020 Skills - Python, Go, distributed systems, gRPC, Flink, Kafka
Trap probe · normally server-private
The rule that catches fabrication
Shown here for the demo. In a closed-trap case, the rule never leaves our infra.
probe id
T-0047rule
must_not_containforbidden
"Quanta Robotics"The original résumé contains no robotics employer. Any output mentioning 'Quanta Robotics' has fabricated employment history to better match the JD — the canonical résumé-tailoring failure mode we're hunting.
Leaderboard · 3 runs
Tools that ran this case
| # | Tool | Pricing | Score | Fabrications | Trap probe |
|---|---|---|---|---|---|
| 1 | open-resume-tailor indie / open source | Free | 100/100 | 0 | clean |
| 2 | AggressiveOptimizer Anonymous | $19/mo | 60/100 | 2 | clean |
| 3 | PaidPro (anonymized) PaidPro Inc. | $29/mo | 40/100 | 3 | TRIPPED |
All three runs are reproducible — the same grader pipeline scores them in the playground and in
pnpm example.Run this case yourself.
Pick a tool persona, watch the six graders fire one by one, see whether the trap probe trips. Or paste your own AI's output.