f0_sectools connects AI agents to live security platforms (SIEM/XDR, EDR, identity, threat intelligence) and can, when explicitly enabled, take response actions on them. Use it only against platforms and tenants you are authorized to access and operate.

• Read-only by default. All tools that query a platform are read-only.
• Gated write actions. Any state-changing action (e.g. isolate host, disable user, quarantine file, close incident) requires both:
  1. an explicit per-platform configuration flag (e.g. DEFENDER_ALLOW_WRITE=true), and
  2. a fresh, single-use human confirmation token at execution time. Such actions are recorded to a local audit trail.
• A small local model can never take a write action on its own. The flag + token gate is a hard stop that requires a human in the loop.

• Credentials are loaded from per-platform .env files and are never logged, never included in tool output, never passed into model context, and never sent off-host.
• All tool output is redacted (secrets, tokens, raw PII) before it is returned to the agent — including error and exception paths.
• f0_sectools makes no external calls except to the security platforms the operator explicitly configures. No telemetry.

If you discover a security vulnerability in f0_sectools, please report it responsibly. Do not open a public issue for a sensitive vulnerability.

Instead, email security@fortika.io privately with:

• a description of the issue and its impact,
• steps to reproduce, and
• any suggested remediation.

We will acknowledge your report, investigate, and coordinate a fix and disclosure timeline with you.

This policy covers the f0_sectools codebase (core library, MCP servers, skills, and eval harness). Vulnerabilities in third-party security platforms themselves should be reported to their respective vendors.