The Agent Skills Directory

[COMMAND_EXECUTION]: The skill executes shell commands to gather project metadata and manage local files within the Obsidian vault.
Evidence: Uses git remote get-url origin to infer project names and the date command for timestamped filenames.
Evidence: Relies on the obsidian CLI utility to read, create, and append content to markdown files.
[INDIRECT_PROMPT_INJECTION]: The skill operates on an attack surface for indirect prompt injection by processing conversation history and performing file write operations based on that content.
Ingestion points: Step 3 instructs the agent to analyze the "full conversation" to extract insights, decisions, and code snippets.
Boundary markers: Absent; there are no delimiters or specific instructions to ignore embedded commands or malicious formatting within the conversation data.
Capability inventory: The skill has filesystem write access through the obsidian create and obsidian append commands.
Sanitization: Absent; the skill does not specify any sanitization, escaping, or validation of the conversation text before it is written to a local file.

til