The Agent Skills Directory

[COMMAND_EXECUTION]: The skill frequently invokes local system commands to manage the OpenSpec workflow. This includes executing openspec status, openspec new change, openspec instructions, and openspec archive, as well as using git log to analyze recent project activity. These commands are integral to the tool's teaching purpose but involve direct shell interaction.
[INDIRECT_PROMPT_INJECTION]: The skill creates an attack surface by ingesting and processing data from the codebase that may be controlled by third parties (e.g., in shared repositories). It specifically scans for TODO, FIXME, and other markers in source files and reads the output of git log to suggest tasks. Malicious instructions placed in these locations could potentially influence the agent's behavior.
Ingestion points: Local source code files and the output of git log (SKILL.md).
Boundary markers: Absent; the skill does not use specific delimiters or instructions to the agent to disregard instructions found within the scanned code comments or commit messages.
Capability inventory: The agent has the capability to write files to the filesystem (specifically within the openspec/ directory) and execute project-related CLI tools.
Sanitization: None; the skill reads and displays code comments and log entries without validation or escaping, which are then used to define the scope of work.

openspec-onboard