Test design, vulnerability assessment, and quality assurance (13K skills)
Check mobile app changes in an iOS simulator using real local services. It covers user interactions, service failures and recovery, language and text size checks, with screenshots and recordings as evidence.
Audit PHP and other source code by listing calls to known dangerous functions, checking for enumeration gaps, and examining input paths and safeguards. Store risk assessments and verdicts in a searchable audit ledger.
Verify a code change by running the app and checking that it works as expected. Use it after a fix or other change to confirm the intended behavior in the running application.
Automate browser tasks such as clicking, filling forms, and switching tabs. Inspect page structure, console messages, and network requests, capture screenshots or videos, and work with Playwright tests to check and investigate web page behavior.
Check which CARLA features are missing on UE 5.5 compared with UE 5.8. Read source markers to identify the checkout, review unsupported workflows, and see which UE 5.8 procedures also apply.
Validate CARLA pedestrian navigation by sampling walkable locations and reporting their coverage. Draw the samples as debug points to inspect the walkable area after a map import or investigate pedestrians that will not move.
Change weather on a running CARLA simulation server using requests such as heavy rain at sunset or a foggy night. The skill maps words to presets or exact parameters, applies them, and reads back the resulting settings.
Evaluate a driving agent on CARLA Leaderboard routes after checking the required setup. Select routes and repetitions, save driving scores and infractions, record runs, and resume interrupted evaluations from a saved checkpoint.
Run CARLA routes with a driving agent through ScenarioRunner, including scenarios available only in route mode. Inspect route contents, execute one or all routes, and obtain reports on completion, collisions, traffic lights, and other criteria.
Summarize ScenarioRunner results to see which scenarios passed and which checks failed. Compare runs before and after changes, or analyze recorded driving to produce measurements and plots of vehicle distance and deviation from the lane center.
Create custom traffic scenarios for CARLA ScenarioRunner with a Python class and XML configuration. Generate a starting skeleton, define actor behavior and pass/fail criteria, and check the files before running them in a simulator.
Run individual or grouped traffic scenarios in CARLA, a driving simulator, and collect pass/fail reports or recordings. Check which scenarios match the current branch and map, and investigate runs that fail or show no movement.
Run Scenic traffic scenarios on a CARLA server. First test whether a valid scene can be generated, then simulate with limits on runs and steps, record the outcome, and restore the world's asynchronous setting afterward.
Read CARLA Leaderboard result files to understand driving scores, route completion, and penalties. Investigate low scores, compare two runs, merge split results, or recalculate scores under version 2.0 or 2.1 rules without rerunning the simulation.
Inspect CARLA driving simulation logs without replaying them. Get recording details, find collisions by participant category, and identify vehicles that barely moved, with times and object IDs to locate moments worth reviewing.
Validate and run OpenSCENARIO traffic scenarios through ScenarioRunner. Check file structure, maps, vehicle definitions, and supported features before simulation. Run both .xosc and .osc files, with global parameter overrides available for OpenSCENARIO 1.x.
Turn a written traffic situation into a Scenic file, which describes possible scenes using placement rules and constraints. Select suitable maps and actors, then check that the file compiles and can generate at least one scene.
Use Meta's Prompt Guard to detect malicious instructions in user prompts and external data. This skill guides input filtering, batch checks for retrieved documents, detection threshold adjustments, and splitting long text for inspection.
Run TVM's TIRx compiler tests alongside tirx-kernels import and correctness checks. It verifies the local build, selects GPUs, runs parallel tests, saves full logs, and guides failure investigation and result reporting.
Helps clarify bug reports written by people, search for duplicate reports, and collect diagnostic evidence. Keeps investigation material separate from a concise issue draft, with the detailed workflow defined in a referenced file.
Verify Riptide changes in a local development stack through browser interactions, the HumanLayer CLI, and database checks. Prepare the environment, exercise specific workflows, and confirm both visible results and stored state.
Scan React code for security, performance, correctness, and architecture issues, with a health score from 0 to 100. Check changes for regressions, prioritize existing problems, and use rule-specific guidance to fix and validate the code.
Test Riptide API workflows locally through the HumanLayer CLI. The skill guides login, test task creation, text uploads, session launches, and result inspection, with troubleshooting for authentication errors, unavailable hosts, and queued sessions.
Test local Stripe webhook handling by forwarding payment and subscription events, checking signatures and response codes, and inspecting logs. Use Test Clocks, which advance simulated time, to verify billing lifecycle behavior in test mode.