The Agent Skills Directory

[COMMAND_EXECUTION]: The skill uses local shell commands to interact with system processes, the Obsidian application, and version control tools.
Evidence: Orchestrates actions using git, gh (GitHub CLI), pgrep, and tasklist. It specifically utilizes the obsidian CLI for app-context tasks such as obsidian read, obsidian open, and obsidian backlinks.
[EXTERNAL_DOWNLOADS]: The skill facilitates the retrieval of data from remote servers to synchronize the local vault.
Evidence: Instructions in SKILL.md and resources/git-sync.md recommend git pull --rebase to fetch updates from user-configured remote repositories.
[DATA_EXFILTRATION]: The skill is designed to transmit vault content to external destinations for backup and multi-device synchronization.
Evidence: The 'Git Sync Mode' defined in SKILL.md and resources/git-sync.md utilizes git push to upload vault data to external Git remotes (typically GitHub).
[INDIRECT_PROMPT_INJECTION]: The skill possesses a classic attack surface where instructions embedded in processed data could attempt to manipulate the agent's actions.
Ingestion points: The agent ingests untrusted data from the local vault via read_note, search_notes, and obsidian read commands specified in SKILL.md and resources/tool-patterns.md.
Boundary markers: The provided instructions do not include specific delimiters or 'ignore' directives to separate user data from system instructions during processing.
Capability inventory: The skill has significant capabilities, including full file system read/write access (write_note, delete_note) and the ability to execute shell commands (git, gh, obsidian).
Sanitization: There is no mention of sanitization or validation routines to filter potential instructions out of the note content before the agent processes them.

obsidian