The Agent Skills Directory

[COMMAND_EXECUTION]: The skill executes shell commands using the git CLI to create safety snapshots before modifying files. It also interacts with the qmd search tool (or a path specified in QMD_CLI) to update and embed the vault index.
[DATA_EXFILTRATION]: The skill accesses the local filesystem to read vault content and configuration files, including .env and ~/.obsidian-wiki/config. While these are used for local processing and no network exfiltration is detected, the skill has high-privilege read access to the user's knowledge base.
[INDIRECT_PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection because it ingests and processes untrusted data from markdown files in the vault.
Ingestion points: All .md files within the Obsidian vault path.
Boundary markers: None identified; content is processed directly for mention detection and relationship inference.
Capability inventory: Subprocess execution (git, qmd), extensive file-write capabilities, and YAML frontmatter modification.
Sanitization: Unicode normalization (NFKD) is performed for character matching, but no content sanitization is implemented to prevent embedded instructions from influencing the relationship inference logic in Step 4c.

cross-linker