The Agent Skills Directory

[CREDENTIALS_UNSAFE]: The skill provides troubleshooting commands that display sensitive API credentials in plain text within the agent's output or terminal logs.
Evidence: The command echo $VASTAI_API_KEY is suggested to verify the environment configuration in SKILL.md.
Evidence: The command env | grep VASTAI is suggested to check local configuration, which can expose secrets to the execution context.
[PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection because it instructs the agent to process and match external data (error logs) that may be controlled by an attacker.
Ingestion points: Error messages and logs provided by the user or external systems in Step 1 of SKILL.md.
Boundary markers: Absent; there are no delimiters or specific instructions to treat the log content as untrusted data.
Capability inventory: The skill has access to Grep and Bash (specifically curl), which could be misused if the agent is tricked by malicious log content.
Sanitization: Absent; the skill lacks any validation or filtering of the input error strings.
[COMMAND_EXECUTION]: The skill utilizes shell commands to probe the local environment, which is a standard diagnostic practice but increases the visibility of system metadata.
Evidence: Use of env and grep to inspect environment variables in SKILL.md.

vastai-common-errors