The Agent Skills Directory

[PROMPT_INJECTION]: No prompt injection or behavior override instructions were found in the skill content.
[DATA_EXFILTRATION]: The skill does not perform any network operations to non-whitelisted domains. It primarily interacts with local markdown files for validation purposes.
[OBFUSCATION]: No obfuscated code, hidden characters, or encoded payloads were detected in the scripts or markdown files.
[REMOTE_CODE_EXECUTION]: The skill includes a local validation script (scripts/okf-bundle-validate.py) which is executed locally by the user or agent. The script performs static analysis and does not download or execute external code.
[INDIRECT_PROMPT_INJECTION]: The skill has an attack surface where it processes untrusted markdown files provided by the user. However, the okf-bundle-validate.py script uses a custom-built, manual parser for frontmatter and content, avoiding unsafe execution environments like eval() or yaml.load(). The capability of the script is limited to reading files and reporting syntax errors, which mitigates the risk of indirect injection triggering dangerous actions.

open-knowledge-format