The Agent Skills Directory

[COMMAND_EXECUTION]: The skill uses subprocess.run() in hot_scanner.py, rumor_scanner.py, and watchlist.py to execute internal scripts and the external bird CLI. These calls are implemented using argument lists rather than shell strings, effectively preventing shell injection vulnerabilities despite processing user-supplied ticker symbols.
[INDIRECT_PROMPT_INJECTION]: The analysis engine ingests untrusted data from external sources including news headlines, Reddit posts, and Twitter/X tweets to calculate sentiment scores. While this creates a surface for indirect prompt injection, it is consistent with the primary purpose of the skill. Developers have implemented some filtering (regex ticker extraction) which mitigates basic exploitation, though explicit boundary markers are not used when presenting this data to the agent.
[EXTERNAL_DOWNLOADS]: Data fetching operations target well-known and reputable financial and news services (Yahoo Finance, CoinGecko, Google News). These downloads are restricted to data retrieval and do not involve remote code execution or script evaluation.
[CREDENTIALS_UNSAFE]: The skill handles Twitter/X authentication tokens by reading from a .env file or environment variables. This is a recommended practice for managing secrets in local development and agent environments, avoiding the risk of hardcoded credentials.

stock-analysis