The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and transform untrusted Odoo modules, which serves as a surface for indirect prompt injection attacks.
Ingestion points: The skill scans and processes Odoo manifest files, XML views, JavaScript, and SCSS files within user-provided project paths (defined in SKILL.md and scripts/utils.py).
Boundary markers: There are no explicit instructions or delimiters defined to prevent the agent from following instructions embedded within the source files of the modules being migrated.
Capability inventory: The skill scripts facilitate broad file system access, including reading, writing, and creating directory backups via shutil.copytree. It also instructs the agent to execute shell commands to test the resulting code.
Sanitization: While the tool performs syntax validation using ast.parse for Python and xml.etree.ElementTree for XML (in scripts/validate.py), it does not verify the semantic safety of the code changes, which could be manipulated if the source module is malicious.
[COMMAND_EXECUTION]: The upgrade workflow requires the agent to execute shell commands to verify the module's compatibility after transformation.
Evidence: The SKILL.md workflow includes a test step instructing the assistant to run python -m odoo -d DB -i MODULE --stop-after-init, which executes the potentially untrusted code generated by the upgrade process.

odoo-upgrade