drive-safety
Skill2 starsChanged 6 months agoMITView on GitHub
Safety guardrails for any change that touches the robot's drive path — UNO firmware motor code, brain motion gateway, HUD drive input, or anything that could affect the wheels-locked default. Refuses dangerous changes and forces a checklist before any drive test.
In this repository3 other files it gives its agents.
- AGENTS.mdAGENTS.md
- .github/copilot-instructions.mdCopilot instructions
- autonomous-devSkill
The file
We split this skill into the rules it gives an agent
4 rules. About 0 are common, 4 are distinctive, and 0 are very specific to this project.
≈0
Common · estimate
Many files like this have a rule of this kind.
≈4
Distinctive · estimate
This author's own choices. Worth a look.
0
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- Distinctive: line 39
Before issuing or facilitating a drive test (wheels actually move), walk through this checklist with the human.
- Distinctive: line 58
After any drive test, automatically prompt the human to re-lock the wheels via:
- Distinctive: line 72
When this skill is active, your responses about the changed code MUST:
- Distinctive: line 79
Don't be silent on safety just because the local change "looks fine." The motion-safety story is reinforced by repetition.
≈ “Common” is an estimate: at least 1 in 5 of the 372,429 skill files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.
Files
Drive safety skill
This skill activates when the planned change touches anything in the motion path. It is a hard veto on dangerous defaults.
When this skill triggers
Activate if the edit set includes any of:
zip-v2/firmware/uno/src/main.cpp(motor/loop code)zip-v2/firmware/uno/include/config.h(timing/safety macros)zip-v2/brain/zip_brain/motion.py(motion gateway)zip-v2/brain/zip_brain/uno_link.py(UART writer)zip-v2/hud/lib/v2/useDriveInput.ts(drive dispatch)zip-v2/hud/components/v2/DrivePanel.tsx(drive UI)- Anything that touches
ZIP_MOTION_LOCKED
What this skill enforces
Hard veto
The skill VETOES (refuses to proceed without explicit human override):
- Any change that sets
ZIP_MOTION_LOCKEDto0as a default in code, systemd drop-in, env example, or docs. - Any change to
SERIAL_BAUDaway from500000. - Any removal of the
STBY=HIGHenforcement in motor writes. - Any removal of the deadman TTL check.
- Any change that lets a non-
motion.pycode path send setpoints to the UNO.
Hard checklist before any drive test
Before issuing or facilitating a drive test (wheels actually move), walk through this checklist with the human. Refuse to proceed if any item is "no":
DRIVE-TEST READINESS:
[ ] Robot is on a propped-up chassis (wheels off the ground)? — first test
[ ] OR robot is on the floor with clear space? — subsequent tests
[ ] Battery voltage > 6.5 V?
[ ] motion.lock_state.locked is true (verify in HUD before unlock)?
[ ] HUD WebSocket is connected?
[ ] UNO is responding to N=120 diagnostics?
[ ] Camera feed is live (so you can see what the robot sees)?
[ ] Human has line-of-sight to the robot?
[ ] STOP key (HUD E-Stop button) is bound and tested?Post-test wheel re-lock
After any drive test, automatically prompt the human to re-lock the wheels via:
ssh zip-jetson "sudo systemctl edit zip-brain" # remove the unlock override
# OR
ssh zip-jetson "sudo systemctl restart zip-brain" # re-applies defaultConfirm motion.lock_state.locked === true in the HUD before declaring
the test complete.
Output discipline
When this skill is active, your responses about the changed code MUST:
- Mention "wheels-locked default" explicitly
- Confirm or call out the three immovable rules
- Suggest the verification chain (build → typecheck → restart → drive-test checklist)
Don't be silent on safety just because the local change "looks fine." The motion-safety story is reinforced by repetition.
Reference
docs/adr/0002-uno-owns-time.mddocs/adr/0003-uart-500k-baud.mddocs/adr/0005-wheels-locked-default.md
Does it work?
Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.
No reviews yet
People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.
- It worked
- Partly
- It didn't
Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.
Your trail
Files you open will collect here, so you can find them again.
Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.
Now do this for your own.
Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.

