Turn a code fix into a review checklist
Using Codex CLI where available: Supply a public patch and authorized code to list related files, concerns and suggested tests.
Security, Science & Hardware · 全民AI / AI for Everyone
Bracket dimensions unclear? Draft a design before building
Give a text AI dimensions and mounting requirements to draft a FreeCAD modeling plan and dimension checks for human review. This is a drafting exercise; the source used Claude Code.
Compare software versions with vulnerability notices
Provide a de-identified software inventory and public advisories. Preserve CVE identifiers, source links and original ratings in a review list.
Proof repair changes the theorem? Check assumptions first
Use Codex with a Rocq test project and proof obligation to prepare a minimal proof repair and checker logs through the test workflow below, then review it. Matching tools and permissions are required.
Simulation differs from measurement? Align waveform conditions
Use Claude Code with previously captured waveform and simulation exports to prepare a difference report with units and conditions through the test workflow below, then review it. Matching tools and permissions are required.
Unsure how to recycle an item? Check local rules
Supply safe household-item photos and local rules to draft material classifications, preparation requirements and questions to verify.
Binary analysis uncertain? Separate evidence from guesses
Give a text AI analysis output from an authorized teaching sample to draft a referenced table of possible functions and routes for human review. This is a drafting exercise; the source used Cursor.
Tenant data might mix? Design isolation tests with fake accounts
Give a text AI fake tenants and a role-access matrix to draft a list of allowed and denied test cases for human review. This is a drafting exercise; the source used Replit Agent.
Printing looks wrong? Compare one setting at a time
Draft a print-test matrix in Claude, then use an already configured Windows 11 printer manually to fill in real paper results.
Plan tests for chat on a limited device
Using Claude where available: Supply simulated memory, network and message limits to draft disconnection, restart and data-retention checks.
When should a robot stop? Review uncertain scenes in simulation
Supply empty simulated scenes or approved test images to draft obstacle, uncertainty and human-takeover checks without operating hardware.
Clarify robot instructions before anything moves
Supply simulated scenes, API rules and coordinate units to draft action preconditions, boundaries and completion evidence without dispatchable commands.
Model distorts after resizing? Map dimension dependencies
Define dimensions and dependencies, then adjust and export a model in CADAM, using OpenSCAD for section checks where needed.
Plan parameter and saving checks for a 3D model
Using Claude where available: Supply test-model parameters, dependencies and saving steps to draft checks for changes and reopening.
What email access is needed? Separate reading, drafting and sending
Supply fictional email tasks and available capabilities to draft a permission table separating reading, drafting, sending and deletion.
Robot-concept exercise: stale observations and unknown units require a stop
Create a stop-reason table from fictional observations, checking for stale state, unknown positions and missing units.