zip
Copilot instructions2 starsChanged 5 months agoMITView on GitHub
This file is read automatically by GitHub Copilot, Cursor, and other IDE-integrated AI agents. It points them at the right context fast. Read AGENTS.md at the repo root first. It's the canonical, comprehensive orientation for any AI agent working in this repo. This file is a short index; AGENTS.md is the source of truth. 1. UNO owns time — motion control lives on the ATmega328P, not the brain or HUD. See ADR 0002. 2. UART is exactly 500000 baud —…
- Installs packages
In this repository3 other files it gives its agents.
- AGENTS.mdAGENTS.md
- autonomous-devSkill
- drive-safetySkill
The file
We split this copilot-instructions.md into the rules it gives an agent
6 rules. About 0 are common, 2 are distinctive, and 4 are very specific to this project.
≈0
Common · estimate
Many files like this have a rule of this kind.
≈2
Distinctive · estimate
This author's own choices. Worth a look.
4
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- Distinctive: line 36
If you find yourself wanting to break one, pause and ask the human.
- Very specific: line 81
C++ (firmware): no dynamic alloc in motor paths.
- Very specific: line 88
Commit node_modules/, .pio/, .venv/, *.engine, *.onnx, *.pt, models/, or anything heavy.
- Very specific: line 93
Use 460800 baud anywhere. It silently transmits at 500000 on AVR.
- Distinctive: line 102
ALWAYS request explicit human review even if CI is green.
- Very specific: line 106
If your agent has persistent memory across sessions, look for these memory files at C:\Users\<user>\.claude\projects\C--Zip\memory\:
≈ “Common” is an estimate: at least 1 in 5 of the 3,779 Copilot instructions 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
GitHub Copilot / Cursor / AI agent instructions
This file is read automatically by GitHub Copilot, Cursor, and other IDE-integrated AI agents. It points them at the right context fast.
TL;DR
Read AGENTS.md at the repo root first. It's the
canonical, comprehensive orientation for any AI agent working in this
repo. This file is a short index; AGENTS.md is the source of truth.
Repo shape (post 2026-06-04 restructure)
jarvis/ Vision-first AI on Jetson (PRIMARY current focus)
depth-lab/ perception-lab/ splat-lab/ llm/
zip-v2/ The autonomous robot
hud/ Next.js 16 cockpit (Windows PC)
brain/ Python service on Jetson (git submodule -> zip-brain)
firmware/ UNO (Arduino) + ESP32-S3 (PlatformIO)
bridge/ Legacy Node.js robot bridge (not in current loop)
zip-v1/ Archived predecessor (read-only)
docs/ Umbrella docs (ARCHITECTURE, ROADMAP, KNOWN_ISSUES, ADRs, ...)
.github/ Repo toolingThe three immovable rules (do not violate)
- UNO owns time — motion control lives on the ATmega328P, not the brain or HUD. See ADR 0002.
- UART is exactly 500000 baud — not 460800. See ADR 0003.
- Wheels locked by default on the bench —
ZIP_MOTION_LOCKED=1is the default. See ADR 0005.
If you find yourself wanting to break one, pause and ask the human.
Where things live now
| You want to edit | Path |
| HUD pages/components | zip-v2/hud/app/, zip-v2/hud/components/v2/ |
| HUD hooks / state / WS | zip-v2/hud/lib/v2/ |
| Brain Python | zip-v2/brain/zip_brain/ (submodule — separate PR) |
| UNO firmware | zip-v2/firmware/uno/src/main.cpp, include/config.h |
| ESP32 camera firmware | zip-v2/firmware/esp32-cam/src/main.cpp |
| Jarvis vision labs | jarvis/depth-lab/, jarvis/perception-lab/, jarvis/splat-lab/ |
| OpenClaw / Qwen exploration | jarvis/llm/ (deferred per ADR 0006) |
| Architecture docs | docs/ARCHITECTURE.md, docs/adr/ |
| Known gotchas | docs/KNOWN_ISSUES.md |
| Phase tracking | docs/ROADMAP.md, zip-v2/docs/PHASES.md |
| Protocols (wire formats) | zip-v2/docs/PROTOCOLS.md |
Common verification commands
# HUD
cd zip-v2/hud
npm install
npm run typecheck
npm run build
# UNO firmware
cd zip-v2/firmware/uno
pio run -e uno
# ESP32 camera firmware
cd zip-v2/firmware/esp32-cam
pio run -e esp32cam_sta
# Brain (on Jetson)
ssh zip-jetson
sudo systemctl restart zip-brain
sudo journalctl -u zip-brain -fCode style highlights
- TypeScript: strict mode; no
anywithout a comment. - Python: type hints on public functions; async-first.
- C++ (firmware): no dynamic alloc in motor paths; guard every
-D-overridable macro with#ifndef.
What you should NEVER do
- Bypass the motion gateway (direct UART writes from a script).
- Change
ZIP_MOTION_LOCKEDdefault to0"for convenience." - Commit
node_modules/,.pio/,.venv/,*.engine,*.onnx,*.pt,models/, or anything heavy. Respect.gitignore. - Force-push to
master. - "Fix"
next lint— it's a known Next.js 16 bug; the placeholder script is intentional. - Use
460800baud anywhere. It silently transmits at 500000 on AVR. See ADR 0003.
When you're done
- Self-review the diff before requesting human review.
- Fill the PR template honestly, especially the "three immovable rules check."
- For safety-critical changes (motion, firmware, brain motion gateway), ALWAYS request explicit human review even if CI is green.
Memory / state
If your agent has persistent memory across sessions, look for these
memory files at C:\Users\<user>\.claude\projects\C--Zip\memory\:
project_zip_v2_session_handoff.md— current statereference_repo_rename_2026_06_04.md— URL historyreference_openclaw_jetson_deploy.md— 8 GB Ollama tuningreference_jetson_perception_deploy.md— YOLO on Jetson recipefeedback_jetson_ssh_use_openssh.md— use stock Windows OpenSSH
These are point-in-time snapshots; verify against the current code before asserting as fact.
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.

