The Agent Skills Directory

[SAFE]: The skill is comprised entirely of Markdown files and text-based reference examples. There are no executable scripts, binaries, or automated installation commands that could perform unauthorized actions on a host system.\n- [PROMPT_INJECTION]: The skill includes defensive prompt engineering instructions in SKILL.md, such as advising the agent to disregard 'ignore previous instructions' markers found in news data. These are security enforcement rules for the agent's behavior and do not constitute an attack.\n- [INDIRECT_PROMPT_INJECTION]: The analysis workflows for financial news acknowledge the risk of malicious instructions embedded in external data. The skill mitigates this by mandating the use of boundary markers and instruction to focus strictly on factual market data.\n- [CREDENTIALS_UNSAFE]: Multiple API request examples in the references directory use the placeholder string 'YOUR_API_KEY' for authorization headers. No real secrets or sensitive credentials are hardcoded in the skill.\n- [EXTERNAL_DOWNLOADS]: The documentation points to legitimate vendor-controlled domains (tradingviewapi.com, tvapis.com) for API console access and integration guides, representing normal vendor functionality.

tradingview-openclaw