AGENT SKILL

run-analysis

Run a StockSage analysis for a ticker through the native engine or explicit Python comparison.

License 许可
Apache-2.0
最近更新
2026-08-16
Stars
★ 12
Forks
0

01 前言 Front Matter

name
run-analysis
examples
[ "analyze AAPL for 2026-05-01", "run StockSage analysis for TSLA on 2026-05-14", "queue analysis for MSFT for 2026-05-01" ]
metadata
{ "allbert.app": "stocksage", "allbert.kind": "native_action", "allbert.actions": "run_analysis", "allbert.version": "0.35.0", "allbert.permissions": "stocksage_analyze", "allbert.confirmation": "required" }
description
Run a StockSage analysis for a ticker through the native engine or explicit Python comparison.
allowed-tools
allbert:action:run_analysis

03 SKILL.md 正文

Workflow

  1. Validate the ticker symbol and ISO-8601 analysis date.
  2. Evaluate :stocksage_analyze through Security Central.
  3. When confirmation is required (the default), create a durable confirmation record and stop. Do not call the bridge yet.
  4. On the approved resume path, run the native StockSage specialist-agent graph by default. Only call the Python bridge when the operator explicitly requested engine: "python" or engine: "both".
  5. Persist the returned analysis in the local StockSage domain tables, including parity diff JSON for explicit engine: "both" requests.
  6. When a queue entry id is provided, update the queue entry status and record a queue run row linking it to the analysis.

Safety

  • Native evidence actions and explicit Python bridge runs may make external market-data API calls. The operator confirmation covers analysis as a whole; evidence calls still preserve Resource Access posture.
  • Raw TradingAgents output is never surfaced in traces, CLI list summaries, or signals. Only bounded structured metadata is shown.
  • :stocksage_analyze has a needs_confirmation safety floor; no setting can lower it to allowed.

04 同仓库更多 Skills