The Agent Skills Directory

[CREDENTIALS_UNSAFE]: The skill reads sensitive credentials from the local filesystem. Specifically, x-search/lib/api.ts accesses ~/.config/env/global.env and ~/.claude/skills/twitter/.env to retrieve X API Bearer tokens and OAuth 1.0a secrets.
[EXTERNAL_DOWNLOADS]: The skill requires installing external tools from third-party sources not included in the trusted vendor list. SKILL.md instructs users to install the bird CLI via brew install steipete/tap/bird or npm install -g @steipete/bird, and to run npx smaug setup for archival features.
[COMMAND_EXECUTION]: The skill executes multiple external commands and tools to function. It uses the bun runtime for its main script, invokes the bird CLI for session-based operations, and uses npx to manage the smaug tool.
[PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection due to its processing of external tweet data.
Ingestion points: Fetches tweet content from the X API via x-search/lib/api.ts.
Boundary markers: Absent; there are no delimiters or instructions to ignore embedded commands within the tweet text.
Capability inventory: The skill can write to the local filesystem (e.g., saving research markdown files and updating watchlist.json in x-search/x-search.ts) and perform network operations such as posting or replying to tweets.
Sanitization: Sanitization is minimal in x-search/lib/format.ts, focusing on display formatting rather than security-focused escaping of potential injection payloads.

twitter