菜单
Implement and maintain the core orchestration and compute pipeline in core/ including base interfaces, engine actors, aggregation, timeframe generation, and indicator generation. Use when changing runtime data flow, worker behavior, or strategy execution mechanics.
用 Codex 或 Claude 帮你安装 复制这段 Prompt，粘贴到 Codex、Claude 或其他助手里，让它检查 Skill 页面并帮你完成安装。
Analyze and evolve the overall architecture of next-gen-algo-trading-bot across ingestion, aggregation, strategy execution, storage, APIs, and UI layers. Use when making cross-module changes, debugging data flow breaks between modules, planning new features, or assessing tradeoffs and risk before refactors.
Implement and maintain broker connectivity for Zerodha in brokers/ including auth, session retry behavior, order/position APIs, and historical data sync helpers. Use when touching login flow, order execution, broker resilience, or market data backfill behavior.
Operate and extend the Typer CLI entrypoints in cli/main.py for engine start/stop and timeframe worker execution. Use when adding commands, wiring new modules into runtime startup, or debugging startup/config/env failures.
Manage and validate runtime configuration schemas and YAML files under config/ plus root config.yaml for the trading engine. Use when adding config fields, fixing env interpolation, or resolving mismatches between configured and supported timeframes/indicators.
Maintain local operator dashboards in dashboard/ using Streamlit and Dash for candle inspection and indicator visualization. Use when improving manual observability, chart behavior, or live diagnostics from parquet-backed data.
Maintain live market data feed adapters in feeds/ with focus on Zerodha websocket integration and normalized tick payloads. Use when adding new feed providers, normalizing tick schema, or debugging live tick ingestion issues.
基于 SOC 职业分类
| name | ngat-core |
| description | Implement and maintain the core orchestration and compute pipeline in core/ including base interfaces, engine actors, aggregation, timeframe generation, and indicator generation. Use when changing runtime data flow, worker behavior, or strategy execution mechanics. |
Drive ingestion-to-signal execution in the Python compute layer. Use this skill to safely modify actor loops, aggregation semantics, and core interface contracts.
core/base.py: abstract interfaces for feed, aggregator, strategy, broker, storage.core/engine.py: Engine, SymbolWorker, OrderAndPositionWorker, Ray orchestration.core/aggregator.py: in-memory tick/candle store and timeframe resampling with parquet dump.core/timeframe_generator_worker.py: aggregate 1m Redis candles into derived frames.core/indicator_generator_worker.py: compute indicators from candles and write signal series.- to +) that do not scale.on_tick flow mostly bypassed by sync workers).indicator_generator_worker timestamp bookkeeping carefully: last_polled mixes millisecond comments with second-based fetches.timeframe_generator_worker data-loop control carefully: break on missing data exits symbol loop early.start, pause, shutdown, heartbeats).