generated: '2026-09-06' method: searched source: >- https://raw.githubusercontent.com/PX4/PX4-Autopilot/main/SECURITY.md (HTTP 200) and https://github.com/PX4/PX4-Autopilot/security/policy (HTTP 200), read 2026-09-06. probe-security-programs.py reported vdp=none because it probes web hosts and /.well-known/security.txt, and this policy lives in the repository — found by hand. provider: Dronecode Foundation providerId: dronecode has_program: true program_type: coordinated-disclosure bug_bounty: false bounty_note: No bug bounty, no HackerOne/Bugcrowd/Intigriti program. Reporters are credited, not paid. policy_url: https://github.com/PX4/PX4-Autopilot/security/policy policy_file: https://github.com/PX4/PX4-Autopilot/blob/main/SECURITY.md scope: project: PX4 Autopilot note: >- The policy covers PX4 Autopilot. MAVSDK, MAVLink and QGroundControl publish NO SECURITY.md — checked 2026-09-06, all three return 404 for /contents/SECURITY.md. There is no Foundation-wide disclosure policy and no /.well-known/security.txt on any host (see well-known/dronecode-well-known.yml — 404 on all eight). intake: channel: GitHub Security Advisories method: >- "Report a Vulnerability" on the Security tab of github.com/PX4/PX4-Autopilot. No email address is published for intake; the fallback is emailing the release managers listed in MAINTAINERS.md. security_txt: false response_commitments: acknowledgement: 7 days quote: 'Acknowledgment: The maintainer team will acknowledge your report within 7 days.' triage: Severity and impact assessed, next steps communicated. disclosure: >- Coordinated with the reporter. "We follow responsible disclosure practices and will credit reporters in the advisory unless they request anonymity." escalation: Follow up by emailing the release managers if no acknowledgement within 7 days. supported_versions: - version: 1.17.x supported: true - version: 1.16.x supported: true - version: < 1.16 supported: false notable: ai_assisted_discovery_policy: >- PX4 publishes an explicit AI-assisted-discovery clause, aligned with the Linux kernel security policy — findings produced with AI assistance must be treated as PUBLIC and submitted as a pull request rather than a private advisory, because the same bug tends to surface across multiple researchers on the same day. The reproducer is the exception and must never be posted publicly: "it works against vehicles in the field until the fix ships." The policy also asks that a reproducer be demonstrated against SITL (`make px4_sitl`) before filing. This is one of the very few provider security policies in this catalog that addresses agent-generated vulnerability reports at all. secure_development_practices: - Peer code review required on all changes. - clang-tidy static analysis on every pull request, warnings treated as errors. - Daily fuzzing pipeline (Google fuzztest) over MAVLink message handling and GNSS driver protocol parsing. - Input validation on all external inputs — MAVLink messages, RC signals, sensor data. - Compiler hardening — -Wall -Werror, stack protectors. gaps: - No /.well-known/security.txt (RFC 9116) on any of the eight hosts probed. - No policy for MAVSDK, MAVLink or QGroundControl — three of the five hosted projects. - No Foundation-level policy at dronecode.org; /security returns 404.