The Agent Skills Directory

[COMMAND_EXECUTION]: The skill relies extensively on the execution of the logfire-cli executable to configure telemetry, gateway routing, and auth status. It instructs the agent to run complex shell commands involving process redirection and argument passing.
[INDIRECT_PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection and potential command injection through the interpolation of untrusted data into shell commands.
Ingestion points: The skill ingests untrusted data in the form of user-provided organization names (--org), project names (--name), and file system paths (--data-dir) in references/connection.md, references/gateway.md, and references/telemetry.md.
Boundary markers: There are no explicit instructions to sanitize or validate these inputs before they are interpolated into the shell commands, although it does warn to use the CLI's native credential storage instead of manual manipulation.
Capability inventory: The agent is granted the capability to execute the logfire-cli tool with arbitrary arguments, including those that modify system-level configurations for other coding agents (e.g., agent gateway and agent claude configure).
Sanitization: The instructions do not specify sanitization or escaping of these parameters before they are passed to the subprocess call, relying on the agent's default behavior.

logfire-coding-agents