The Agent Skills Directory

[REMOTE_CODE_EXECUTION]: The skill's execution workflow instructs the agent to run existing shell scripts found within the target repository path, specifically PROJECT_PATH/OpenTelemetry/install_opentelemetry.sh and run_codex_with_otel.sh. Because these scripts originate from the project being analyzed (which may be untrusted), this facilitates the execution of unverified third-party code.\n- [INDIRECT_PROMPT_INJECTION]: The skill is susceptible to indirect instructions embedded in the target project. \n
Ingestion points: The skill reads and interacts with the file system at the user-provided PROJECT_PATH.\n
Boundary markers: No delimiters or protective instructions are utilized when processing the repository content.\n
Capability inventory: The skill possesses capabilities to execute shell scripts, write to configuration files (config.toml), and run system detection commands.\n
Sanitization: There is no evidence of validation or sanitization for the scripts or configuration files found within the project path before they are executed or parsed.\n- [DATA_EXFILTRATION]: The skill configures OpenTelemetry to export telemetry data to a user-defined OTLP_ENDPOINT. It explicitly enables log_user_prompt = true, which transmits the user's interaction history and prompt content to an external service. While the skill includes guidance on redacting secrets, the default configuration involves the transmission of potentially sensitive prompt data.\n- [COMMAND_EXECUTION]: The skill performs various shell operations to detect runtimes and manage dependencies, which are executed within the environment of the user-provided PROJECT_PATH.

opentelemetry-codex