hailo-apps / rules
Cursor rule506 starsChanged 7 months agoMITView on GitHub
All skills, instructions, toolsets, knowledge bases, and memory live in .hailo/. Read .hailo/README.md for the complete master index. Always walk through key decisions with the user before building. Ask 2-3 targeted questions to confirm app type, features, and input source. This creates a collaborative workflow and catches misunderstandings early. Only skip questions if the user explicitly says "just build it" or "use defaults". Based on what the task involves, read only the matching rows: Persistent knowledge in .hailo/memory/. Read at…
We split this hailo-global.mdc into the rules it gives an agent
9 rules. About 0 are common, 0 are distinctive, and 9 are very specific to this project.
≈0
Common · estimate
Many files like this have a rule of this kind.
≈0
Distinctive · estimate
This author's own choices. Worth a look.
9
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- Very specific: line 7
All skills, instructions, toolsets, knowledge bases, and memory live in .hailo/.
- Very specific: line 8
Read .hailo/README.md for the complete master index.
- Very specific: line 20
HEF resolution: Always use resolve_hef_path(path, app_name, arch)
- Very specific: line 21
Device sharing: Always use SHARED_VDEVICE_GROUP_ID when creating VDevice
- Very specific: line 23
CLI parsers: Use get_pipeline_parser() or get_standalone_parser()
- Very specific: line 24
Architecture detection: Use detect_hailo_arch() or --arch flag
- Very specific: line 25
USB camera: Always --input usb for auto-detection.
- Very specific: line 26
SKILL.md is sufficient: Read SKILL.md + common_pitfalls.md only.
- Very specific: line 27
Custom background: Use background.copy() — never blend camera feed with background.
≈ “Common” is an estimate: at least 1 in 5 of the 6,350 Cursor rule 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
--- alwaysApply: true --- # Hailo Apps — Global Rules All skills, instructions, toolsets, knowledge bases, and memory live in `.hailo/`. Read `.hailo/README.md` for the complete master index. ## Interactive Workflow (MUST FOLLOW) **Always walk through key decisions with the user before building.** Ask 2-3 targeted questions to confirm app type, features, and input source. This creates a collaborative workflow and catches misunderstandings early. Only skip questions if the user explicitly says "just build it" or "use defaults". ## Critical Conventions 1. **Imports are always absolute**: `from hailo_apps.python.core.common.xyz import ...` 2. **HEF resolution**: Always use `resolve_hef_path(path, app_name, arch)` 3. **Device sharing**: Always use `SHARED_VDEVICE_GROUP_ID` when creating VDevice 4. **Logging**: Use `get_logger(__name__)` 5. **CLI parsers**: Use `get_pipeline_parser()` or `get_standalone_parser()` 6. **Architecture detection**: Use `detect_hailo_arch()` or `--arch` flag 7. **USB camera**: Always `--input usb` for auto-detection. Never hardcode `/dev/video0`. 8. **SKILL.md is sufficient**: Read SKILL.md + common_pitfalls.md only. Do NOT read source code. 9. **Custom background**: Use `background.copy()` — never blend camera feed with background. ### Context Routing Table Based on what the task involves, read **only** the matching rows: | If the task mentions... | Read these files | |---|---| | **VLM, vision, image understanding** | `.hailo/skills/hl-build-vlm-app.md`, `.hailo/toolsets/vlm-backend-api.md`, `.hailo/memory/gen_ai_patterns.md` | | **LLM, chat, text generation** | `.hailo/skills/hl-build-llm-app.md`, `.hailo/instructions/gen-ai-development.md`, `.hailo/toolsets/gen-ai-utilities.md`, `.hailo/memory/gen_ai_patterns.md` | | **Agent, tools, function calling** | `.hailo/skills/hl-build-agent-app.md`, `.hailo/toolsets/gen-ai-utilities.md`, `.hailo/memory/gen_ai_patterns.md` | | **Voice, STT, TTS, Whisper, speech** | `.hailo/skills/hl-build-voice-app.md`, `.hailo/toolsets/gen-ai-utilities.md` | | **Pipeline, GStreamer, video, stream** | `.hailo/skills/hl-build-pipeline-app.md`, `.hailo/instructions/gstreamer-pipelines.md`, `.hailo/toolsets/gstreamer-elements.md`, `.hailo/memory/pipeline_optimization.md` | | **Game, interactive, pose game** | `.hailo/skills/hl-build-pipeline-app.md`, `.hailo/toolsets/pose-keypoints.md`, `.hailo/toolsets/core-framework-api.md`, `.hailo/memory/common_pitfalls.md` | | **Standalone, OpenCV, HailoInfer** | `.hailo/skills/hl-build-standalone-app.md`, `.hailo/toolsets/core-framework-api.md` | | **Camera, USB, RPi, capture** | `.hailo/skills/hl-camera.md`, `.hailo/memory/camera_and_display.md` | | **HEF, model, download, config** | `.hailo/skills/hl-model-management.md`, `.hailo/toolsets/hailort-api.md`, `.hailo/memory/hailo_platform_api.md` | | **Monitoring, events, alerts** | `.hailo/skills/hl-monitoring.md`, `.hailo/skills/hl-event-detection.md` | | **Testing, validation, pytest** | `.hailo/skills/hl-validate.md`, `.hailo/instructions/testing-patterns.md` | | **ALWAYS read (every task)** | `.hailo/memory/common_pitfalls.md`, `.hailo/instructions/coding-standards.md` | ## Available Skills | Skill | Doc | |-------|-----| | Build VLM App | `.hailo/skills/hl-build-vlm-app.md` | | Build Pipeline App | `.hailo/skills/hl-build-pipeline-app.md` | | Build Standalone App | `.hailo/skills/hl-build-standalone-app.md` | | Build Agent App | `.hailo/skills/hl-build-agent-app.md` | | Build LLM App | `.hailo/skills/hl-build-llm-app.md` | | Build Voice App | `.hailo/skills/hl-build-voice-app.md` | ## Memory Persistent knowledge in `.hailo/memory/`. Read at task start, update when learning.
Where files are headingNew Cursor rule files have more rules about planning and tracking tasks, and fewer about code style.
Of the Cursor rule files we read, the 3,978 first seen in the last six months, against the older ones: the kind of rule that gained most share, and the kind that lost most. About files like this in general, not this one, and an estimate, since a kind is a pattern in the words.
In this repository65 other files it gives its agents.
- CLAUDE.mdCLAUDE.md
- .github/copilot-instructions.mdCopilot instructions
- .cursor/rules/core-framework.mdcCursor rule
- .cursor/rules/gen-ai-apps.mdcCursor rule
- .cursor/rules/hl-build-agent-app.mdcCursor rule
- .cursor/rules/hl-build-llm-app.mdcCursor rule
- .cursor/rules/hl-build-pipeline-app.mdcCursor rule
- .cursor/rules/hl-build-standalone-app.mdcCursor rule
- .cursor/rules/hl-build-vlm-app.mdcCursor rule
- .cursor/rules/hl-build-voice-app.mdcCursor rule
- .cursor/rules/hl-camera.mdcCursor rule
- .cursor/rules/hl-event-detection.mdcCursor rule
- .cursor/rules/hl-model-management.mdcCursor rule
- .cursor/rules/hl-monitoring.mdcCursor rule
- .cursor/rules/hl-plan-and-execute.mdcCursor rule
- .cursor/rules/hl-validate.mdcCursor rule
- .cursor/rules/inst-agent-protocols.mdcCursor rule
- .cursor/rules/inst-architecture.mdcCursor rule
- .cursor/rules/inst-coding-standards.mdcCursor rule
- .cursor/rules/inst-gen-ai-development.mdcCursor rule
- .cursor/rules/inst-gstreamer-pipelines.mdcCursor rule
- .cursor/rules/inst-orchestration.mdcCursor rule
- .cursor/rules/inst-testing-patterns.mdcCursor rule
- .cursor/rules/memory-camera_and_display.mdcCursor rule
- .cursor/rules/memory-common_pitfalls.mdcCursor rule
- .cursor/rules/memory-gen_ai_patterns.mdcCursor rule
- .cursor/rules/memory-hailo_platform_api.mdcCursor rule
- .cursor/rules/memory-MEMORY.mdcCursor rule
- .cursor/rules/memory-pipeline_optimization.mdcCursor rule
- .cursor/rules/pipeline-apps.mdcCursor rule
- .cursor/rules/prompt-new-agent-tool.mdcCursor rule
- .cursor/rules/prompt-new-llm-app.mdcCursor rule
- .cursor/rules/prompt-new-pipeline-app.mdcCursor rule
- .cursor/rules/prompt-new-pose-game.mdcCursor rule
- .cursor/rules/prompt-new-standalone-app.mdcCursor rule
- .cursor/rules/prompt-new-vlm-variant.mdcCursor rule
- .cursor/rules/prompt-new-voice-app.mdcCursor rule
- .cursor/rules/prompt-orchestrated-build.mdcCursor rule
- .cursor/rules/standalone-apps.mdcCursor rule
- .cursor/rules/tests.mdcCursor rule
- .cursor/rules/toolset-core-framework-api.mdcCursor rule
- .cursor/rules/toolset-gen-ai-utilities.mdcCursor rule
- .cursor/rules/toolset-gstreamer-elements.mdcCursor rule
- .cursor/rules/toolset-hailort-api.mdcCursor rule
- .cursor/rules/toolset-pose-keypoints.mdcCursor rule
- .cursor/rules/toolset-vlm-backend-api.mdcCursor rule
- .cursor/rules/toolset-yolo-coco-classes.mdcCursor rule
- hl-build-agent-appSkill
- hl-build-llm-appSkill
- hl-build-pipeline-appSkill
- hl-build-standalone-appSkill
- hl-build-vlm-appSkill
- hl-build-voice-appSkill
- hl-cameraSkill
- hl-event-detectionSkill
- hl-model-managementSkill
- hl-monitoringSkill
- hl-plan-and-executeSkill
- hl-validateSkill
- hl-build-agent-appSkill
- The first 60 shown.
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.