The Agent Skills Directory

[COMMAND_EXECUTION]: The skill uses the Bash tool to execute several internal Python scripts for processing data and managing the wiki. It correctly implements security best practices by passing untrusted metadata (such as titles and summaries) through environment variables rather than interpolating them directly into shell commands, effectively preventing command injection.
[EXTERNAL_DOWNLOADS]: The skill uses the WebFetch tool to retrieve content from user-provided URLs. This is a core part of its intended functionality for web page and PDF ingestion.
[INDIRECT_PROMPT_INJECTION]: The skill ingests untrusted data from the internet and local files, which is a known vector for indirect prompt injection attacks.
Ingestion points: Data enters the system via the --url, --file, --paste, and --interview parameters in SKILL.md.
Boundary markers: While specific delimiters are not detailed in this orchestrator, the skill delegates actual content extraction to a separate source-ingester agent.
Capability inventory: The skill utilizes Bash, Write, and Task tools to perform its duties.
Sanitization: The workflow includes specific steps to sanitize summaries and uses robust environment variable passing to prevent untrusted content from influencing command execution.

knowledge-ingest-source