The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill instructions refer to a script file src/communityNotes.js which is not included in the skill package. Consequently, the logic and safety of this dependency cannot be audited or verified.
[DYNAMIC_EXECUTION]: The user is instructed to paste JavaScript code from src/communityNotes.js directly into the browser DevTools console. This execution method bypasses browser security controls and standard skill execution sandboxes, presenting a risk of account manipulation or data access if the script is untrustworthy.
[INDIRECT_PROMPT_INJECTION]: The skill processes data from external X/Twitter posts and community notes.
Ingestion points: Browser DOM content containing tweets and fact-check notes (SKILL.md).
Boundary markers: None; there are no delimiters or instructions to ignore embedded commands in the processed data.
Capability inventory: Reading DOM content and performing automated actions like writing or rating notes (SKILL.md).
Sanitization: No sanitization or content filtering is implemented for the data read from the browser.
[METADATA_POISONING]: There is a mismatch between the provided author context (nirholas) and the author field in the skill's YAML metadata (nichxbt). This discrepancy may indicate an attempt to misrepresent the skill's origin or provide misleading information about its maintainer.

community-notes