The Agent Skills Directory

[COMMAND_EXECUTION]: The skill instructs the agent to wrap various tasks in the starbridge run command, enabling the execution of arbitrary shell commands through the CLI tool's reporting wrapper.
[PRIVILEGE_ESCALATION]: The instructions guide the agent to request permission to run commands outside the sandbox environment (specifically mentioned for Codex) to facilitate network access required by the tool.
[INDIRECT_PROMPT_INJECTION]: The skill defines a process for receiving and acting upon user input from an external device (mobile phone), which is then fed back into the agent's session prompt. 1. Ingestion points: User responses to mobile cards, arriving as session prompts (described in SKILL.md). 2. Boundary markers: No explicit boundary markers or 'ignore' instructions are provided for interpreting the user's mobile response. 3. Capability inventory: The agent is capable of shell command execution and environment manipulation as described in the skill. 4. Sanitization: No sanitization or validation rules are specified for free-form user replies received through the tool.

starbridge