firecrawl-knowledge-base
Pass
Audited by Gen Agent Trust Hub on Aug 21, 2026
Risk Level: SAFE
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and process large amounts of untrusted external data from the web to create knowledge bases. This creates an indirect prompt injection surface where instructions embedded in scraped web pages could influence the agent's behavior.
- Ingestion points: Web content retrieved via Firecrawl map, search, and scrape commands (SKILL.md).
- Boundary markers: The skill instructions do not explicitly mandate the use of delimiters or 'ignore instructions' markers when processing the retrieved markdown.
- Capability inventory: The skill involves writing markdown files, JSON manifests, and JSONL training data to the local file system (SKILL.md).
- Sanitization: No explicit sanitization steps for embedded instructions are defined in the workflow instructions.
- [CREDENTIALS_SAFE]: The skill requires a
FIRECRAWL_API_KEY. This is requested as a standard input field, which is consistent with secure secret management practices for agent skills. No hardcoded credentials were found. - [COMMAND_EXECUTION]: The skill suggests the use of sub-agents or parallel task runners for scraping and chunk generation. While this involves agent orchestration, no arbitrary shell command execution patterns were detected.
Audit Metadata