The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: Automated security scans identified a package URL from the official NPM registry (https://registry.npmjs.org/@esbuild/win32-ia32/-/win32-ia32-0.27.7.tgz). This is a standard dependency for the 'esbuild' build tool and is considered a safe resource from a well-known service.
[PROMPT_INJECTION]: The skill possesses an indirect prompt injection surface as it retrieves untrusted data from the X platform (tweets, user profiles, mentions) and processes it within the agent's context.
Ingestion points: Data enters the agent context through API responses in src/lib/api.ts (e.g., getTweet, searchTweets, getTimeline).
Boundary markers: There are no explicit instructions or delimiters in the skill scripts to warn the agent about potential instructions embedded in tweet content.
Capability inventory: The skill allows the agent to perform write operations like tweet post, tweet delete, and like if the user has explicitly enabled read-write mode.
Sanitization: No specific sanitization or filtering is performed on retrieved social media content before it is displayed to the agent.
[SAFE]: The skill implements several security best practices:
Credential Isolation: The auth export command, which reveals sensitive OAuth tokens, is programmatically disabled when the tool is running as an MCP server (src/index.ts). This prevents the agent from exfiltrating its own authentication state.
File System Security: Configuration and token files are created with restricted 'owner-only' permissions (mode 0o600) to prevent unauthorized local access (src/lib/private-files.ts).
Least Privilege: The tool defaults to read-only mode and requires explicit user action to enable write capabilities.

x-cli