The Agent Skills Directory

[COMMAND_EXECUTION]: The skill is designed to execute shell commands using the qmd CLI (or a path provided by the QMD_CLI environment variable). These commands (update, embed, ls, get) are used to refresh search vectors and verify index integrity after the skill writes new digest files to the vault.
[INDIRECT_PROMPT_INJECTION]: The skill ingests and summarizes the content of all Markdown files within the vault that were active during a specified period. It lacks explicit instructions to treat this content as untrusted or to use boundary markers, which could allow malicious instructions embedded in a wiki page to influence the agent's summary or behavior.
Ingestion points: Reads all .md files under $OBSIDIAN_VAULT_PATH, as well as log.md, hot.md, and _insights.md (SKILL.md, Step 1 and 4).
Boundary markers: None identified in the instructions for reading or summarizing vault content.
Capability inventory: Includes file writing (journal digests, log updates, index updates) and shell command execution (qmd commands in Step 7).
Sanitization: No explicit sanitization, filtering, or validation is performed on the Markdown content before it is processed for themes, connections, or headlines.

wiki-digest