The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill ingests untrusted markdown data from an Obsidian vault (specifically wiki/log.md and referenced child pages) to generate summaries and rollup content. This data could contain malicious instructions intended to manipulate the agent's behavior during the summary generation or the subsequent transaction process.
Ingestion points: Processes wiki/log.md and various child pages within the user-specified vault.
Boundary markers: No specific delimiters or "ignore previous instructions" markers are used when processing the content of the log entries.
Capability inventory: The skill uses python3 to execute a local script that writes changes to wiki/folds/, wiki/index.md, and wiki/log.md.
Sanitization: While the instructions emphasize "extractive summary," there is no explicit sanitization or validation performed on the ingested markdown content to prevent prompt injection attacks.
[COMMAND_EXECUTION]: The skill executes shell commands to interact with a core Python utility script for vault management.
Evidence: The SKILL.md file explicitly instructs the agent to run python3 "$CORE" transaction inspect, python3 "$CORE" transaction apply, and python3 "$CORE" checkpoint to manage file writes and version control.

wiki-fold