The Agent Skills Directory

[COMMAND_EXECUTION]: The skill instructions direct the agent to construct and execute shell commands that incorporate raw user input. Specifically, the command no-mistakes axi run --intent "<intent>" is constructed using text provided by the user. The skill explicitly tells the agent to pass this text verbatim, creating a command injection vulnerability where a user can execute arbitrary shell commands by including characters like backticks, semicolons, or unmatched quotes in their task description.
[PROMPT_INJECTION]: The skill is susceptible to prompt injection by design, as it instructs the agent to take arbitrary user task descriptions and use them to drive the logic of an external tool without any validation or filtering.
[INDIRECT_PROMPT_INJECTION]: The skill exposes a significant attack surface for indirect prompt injection through its processing of untrusted user tasks.
Ingestion points: User-provided task text is captured via the /no-mistakes command and task descriptions in SKILL.md.
Boundary markers: Absent. The skill provides no instructions for using delimiters or warnings to ignore instructions embedded within the user's input.
Capability inventory: The agent can modify local files, execute the no-mistakes CLI, and perform remote git operations (push, PR creation).
Sanitization: Absent. The agent is explicitly told to pass user input verbatim into the validation pipeline.
[EXTERNAL_DOWNLOADS]: The skill relies on an external CLI tool called no-mistakes and references the domain toonformat.dev. The instructions for no-mistakes init and no-mistakes doctor indicate that the skill operates by executing this non-standard software, whose provenance and integrity are not verified within the skill context.

no-mistakes