The Agent Skills Directory

[PRIVILEGE_ESCALATION]: The skill's documentation explicitly instructs the use of sudo apt-get install wkhtmltopdf for Linux environments. An AI agent with bash capabilities may attempt to execute these commands, bypassing standard permission constraints.
[CREDENTIALS_UNSAFE]: The skill targets odoo.conf for configuration tasks. In Odoo environments, this file is the primary repository for sensitive information, including database passwords (db_password) and the master system password (admin_passwd). Accessing or modifying this file presents a significant risk of credential exposure.
[COMMAND_EXECUTION]: The skill provides various shell commands for package installation (apt-get, brew, winget) and environment setup. While intended as guidance, these represent potential execution paths for an agent with shell access.
[INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and transform untrusted data in the form of Odoo XML templates and Python manifest files.
Ingestion points: File patterns in the metadata target **/data/*.xml and **/__manifest__.py across the project directory.
Boundary markers: No specific delimiters or "ignore instructions" markers are defined for the content being processed.
Capability inventory: The skill utilizes Bash, Read, Write, Edit, and WebFetch, creating a wide surface for malicious instructions within templates to trigger external actions.
Sanitization: The skill performs syntax and version validation but lacks mechanisms to detect or sanitize embedded malicious prompts or logic within the QWeb/Jinja2 templates.

odoo-report