The Agent Skills Directory

[CREDENTIALS_UNSAFE]: The wrapper script scripts/run_search.sh reads and parses ~/.bashrc and ~/.zshrc to extract the TWITTER_API_KEY. Accessing these files is risky as they often contain other sensitive credentials and environment variables.
[COMMAND_EXECUTION]: In scripts/run_search.sh, the script uses eval on the output of a grep command targeting shell configuration files. This pattern can lead to arbitrary command execution if the configuration file contains maliciously crafted lines.
[EXTERNAL_DOWNLOADS]: The shell script scripts/run_search.sh automatically triggers pip3 install requests --user if the dependency is missing, which constitutes an unverified runtime download of external code.
[DATA_EXFILTRATION]: The Python script scripts/twitter_search.py transmits the user's API key and search parameters to a non-whitelisted third-party domain https://api.twitterapi.io.
[PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection because it processes untrusted tweet content from an external source.
Ingestion points: Data enters the system via scripts/twitter_search.py which fetches tweet text from the Twitter API.
Boundary markers: The instructions in SKILL.md do not specify the use of delimiters or 'ignore' instructions for the agent when processing the fetched tweet text.
Capability inventory: The skill has the capability to execute shell commands and perform network operations via scripts/run_search.sh and scripts/twitter_search.py.
Sanitization: There is no evidence of sanitization or filtering of the tweet content before it is passed to the agent for analysis and report generation.

twitter-search