The Agent Skills Directory

[COMMAND_EXECUTION]: The skill executes arbitrary shell commands defined within .rota/qa/*.md strategy files during the 'run' mode. While this is the intended function of a QA tool, it creates a risk of code execution if the configuration files are modified by an attacker or if the agent is misled during the scaffolding process.
[INDIRECT_PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection through project data.
Ingestion points: The skill ingests untrusted data from the local project (source code, configuration files like package.json) during the first-run phase and the 'running surface' (live application) during the Audit Pass phase.
Boundary markers: There are no explicit boundary markers or instructions to the LLM to ignore potentially malicious instructions embedded in the audited artifacts.
Capability inventory: The skill can execute shell commands, perform network requests via curl, and write files to the local repository.
Sanitization: The skill lacks sanitization mechanisms for content extracted from the project environment before using it in prompts or strategy generation.
[EXTERNAL_DOWNLOADS]: The skill uses curl to perform network probes against development and staging URLs to verify infrastructure readiness before running tests.

rota-qa