The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill instructs the agent to read and analyze runtime logs collected from the user's application using npx @dev-log/cli logs. If the application logs data that originates from untrusted external sources (e.g., user input, API responses), an attacker could embed malicious instructions in those logs to influence the agent's subsequent analysis or actions.
Ingestion points: SKILL.md directs the agent to read logs via the logs command.
Boundary markers: There are no specified delimiters or instructions for the agent to ignore embedded commands within the log output.
Capability inventory: The agent has the capability to execute shell commands, write files (instrumentation), and perform network operations via the included server and tunnel commands.
Sanitization: No explicit sanitization or filtering of the collected log content is performed before it is presented to the agent for analysis.
[COMMAND_EXECUTION]: The CLI utility uses child_process.spawn to manage a background daemon process for the log server. It also frequently executes its own binary via npx to perform operations like starting the server, generating snippets, and reading logs.
[EXTERNAL_DOWNLOADS]: The skill utilizes the localtunnel package to create HTTPS tunnels, enabling remote access to the local log server. This is a well-known service used for development and debugging purposes.
[DYNAMIC_EXECUTION]: The gen command dynamically constructs code snippets for 13 different programming languages based on provided arguments. These snippets are designed to be inserted into the user's source code by the agent to perform HTTP-based logging at runtime.
[DATA_EXPOSURE]: Collected logs are persisted in a JSON file within the system's temporary directory (os.tmpdir()/dev-log/dev-logs.json). The skill documentation proactively warns against logging sensitive information and provides guidance on stripping credentials from log payloads.

dev-log