The Agent Skills Directory

[COMMAND_EXECUTION]: The skill makes extensive use of the nitro agent CLI tool to manage tasks, memory, and internal communication. These commands are integral to the skill's functionality and include operations for creating tasks, registering agent roles, and sending mail between agents. The instructions also involve standard utility commands such as git for version control and jq for processing JSON output from the CLI.
[INDIRECT_PROMPT_INJECTION]: The skill operates on untrusted data from multiple sources, creating an attack surface for indirect prompt injection.
Ingestion points: Untrusted content enters the agent's context through user-provided "ideas" for projects and findings.md files generated by the nitro-research subagent skill.
Boundary markers: The skill documentation explicitly recommends using quoted heredocs (<<'EOF') when passing multi-line strings to shell commands to prevent unintended shell expansion and improve command safety.
Capability inventory: The skill utilizes shell commands via nitro agent, performs file system operations (creating design documents and research folders), and communicates with other agents via a mail system.
Sanitization: While the skill uses secure command construction patterns, the synthesized content is incorporated into implementation tasks and design documents that may influence future agent actions.
[EXTERNAL_DOWNLOADS]: The skill provides a link to the official installation documentation for the nitro CLI tool at https://chillicream.com/docs/nitro/cli/installation. This is a vendor-owned resource used for legitimate environment setup and does not represent an unauthorized download.

nitro-wayfinder