What problem does it solve? Firecrawl operations can leak credentials, fetch unsafe URLs, mutate account state, or exfiltrate private files when agents act without guardrails. This Skill supplies a mandatory supplemental policy that constrains every firecrawl-* core, build, or workflow skill with strict local safety gates. ## Core Features & Use Cases - Fail-closed URL gate: Validates every URL through literal argv, HTTPS-only checks, DNS resolution, and redirect trust boundaries before any fetch. - Consent and egress controls: Requires action-level consent for browser interactions and explicit approval before parsing sensitive local files through Firecrawl. - Secret and evidence handling: Blocks API keys in argv, enforces 0700/0600 evidence permissions, and mandates scrubbing of cookies, tokens, and PII. - Use Case: When an agent is asked to scrape a customer URL or parse a confidential PDF, this policy forces URL validation, egress disclosure, and approval checks before any command runs. ## Quick Start Load the firecrawl-policy skill alongside any firecrawl skill so its safety gates apply to every Firecrawl operation.
