Agent skill
Create New Indicator
Use this skill when you need to add a new Python indicator to AERA.
Install this agent skill to your Project
npx add-skill https://github.com/CheekyCodexConjurer/quant-lab/tree/main/skills/create_new_indicator
SKILL.md
Create New Indicator (Python)
Use when adding a new indicator under the Strategy Lab indicators/ workspace.
Constraints (platform rules)
- Do NOT hardcode indicator id/name/kind in the frontend renderer.
- Keep the output generic via Plot API (v1 today).
Steps
- Create the script file
- Path:
server/indicators/<name>.py(or subfolders underserver/indicators/). - Must export
calculate(inputs, settings=None). - Optional:
initialize(ctx)to declare Inputs UI (Manifest).
- Produce Plot API v1 output (recommended)
- Return
{"plots": [...]}. - Prefer index-based coordinates (
index,from/to) so the backend can align to candletime. - Reference:
docs/public/plot-api-v1.md.
- No manual registration
- The backend scans
server/indicators/automatically:server/src/services/indicatorFileService.jsserver/src/indicatorRegistry/registry.js
- The indicator id is the relative path including
.py(example:my_indicator.pyormarket_structure/pivots.py).
- Verify
- In the app: Strategy Lab -> save the file; Chart -> activate and run.
- Optional quick check: Debug -> terminal command
run indicator <id> --asset=CL1! --tf=M15 --len=1000. - Run standard validations: see
skills/verify_changes/SKILL.md.
Template
Start from skills/create_new_indicator/template_indicator.py.
Recommended Agent Skills
Expand your agent's capabilities with these related and highly-rated skills.
action_logger
Keep an audit trail of changes, commands, and verification.
size_guard
Enforce line budgets and rotate or split large context files.
analyze_repo_capabilities
Detect repo tooling and patterns, then record a dynamic manifest.
safety_validation
Validate changes against .agentignore before commit.
safe_refactor_migration
Plan and execute large refactors with shadow mode and rollback.
command_guard
Enforce .agentpolicy rules before executing commands.
Didn't find tool you were looking for?