The Agent Skills Directory

[COMMAND_EXECUTION]: The skill includes two local shell scripts (validators/pre-check.sh and validators/post-check.sh) designed for CI validation. These scripts perform benign operations such as checking environment variables and searching for text patterns in output to ensure safety constraints are met. They do not execute arbitrary user input or perform dangerous system operations.
[DATA_EXFILTRATION]: No network operations or external data transmissions were detected. The skill's operations are confined to the local environment and the Obsidian vault.
[PROMPT_INJECTION]: The instructions follow standard operational guidelines and do not contain any attempts to bypass safety filters, override system prompts, or extract internal instructions.
[REMOTE_CODE_EXECUTION]: There are no remote code downloads, external package installations, or dynamic execution patterns. The skill relies on locally defined logic and trusted platform tools like obsidian-mcp.
[SAFE]: The skill explicitly defines 'Gotchas' and 'Constraints' to prevent common agent failure modes, such as unauthorized configuration editing or excessive token consumption through vault-wide scanning. It implements a 'defer to human' policy for any potentially breaking changes.

opencode-context-maintenance