The Agent Skills Directory

[PROMPT_INJECTION]: The skill relies on an indirect prompt injection surface by instructing the agent to ingest runtime data from an Odoo instance to inform its code generation and modification tasks.
Ingestion points: External data enters the agent's context through tools like odoo-ai all <model> and odoo-ai surface, which retrieve JSON briefs containing field definitions, MRO, and view architectures (SKILL.md).
Boundary markers: The instructions lack explicit boundary markers or warnings directing the agent to ignore instructions that might be embedded within the retrieved introspection data.
Capability inventory: The suite provides the agent with the ability to write and modify Python code (odoo-dev), XML views (odoo-views), and JavaScript components (odoo-owl).
Sanitization: There is no mention of sanitizing or validating the retrieved Odoo metadata before it is processed by the AI to generate code.
[COMMAND_EXECUTION]: The skill explicitly directs the agent to execute various commands using the odoo-ai CLI tool for discovery and introspection.
Evidence: The instructions include commands such as odoo-ai native-check, odoo-ai capabilities, odoo-ai all, odoo-ai surface, and odoo-ai esg.

odoo