generated: '2026-08-27' method: searched source: https://openbao.org/community/policies/cve/ ; https://github.com/openbao/openbao/security/policy ; https://github.com/openbao/openbao/security/advisories ; https://github.com/openbao/openbao/blob/main/.github/security-insights.yml name: OpenBao vulnerability disclosure published: true note: >- The automated probe (probe-security-programs.py) found nothing, because it looks for a /.well-known/security.txt and a vendor security page on the web domain and OpenBao has neither - openbao.org is a static Docusaurus site on GitHub Pages and returns 404 for every /.well-known path. The program itself is nevertheless real, published and unusually detailed. It lives on the community policies pages and in GitHub, which is where an open-source project's disclosure process belongs. This artifact records what was actually found there. contacts: - kind: mailing-list value: openbao-security@lists.openssf.org preferred: true note: >- The preferred channel, especially for embargoed issues. Includes all active voting Dev WG members who opted into the security patching process, plus Linux Foundation representatives for oversight. - kind: mailing-list value: openbao@lists.openssf.org preferred: false note: General project list, not the security list. - kind: github-private-report value: https://github.com/openbao/openbao/security/advisories/new note: GitHub private vulnerability reporting is enabled on the repository. policy_urls: - https://openbao.org/community/policies/cve/ - https://github.com/openbao/openbao/security/policy process: notification_channels: - The security mailing list (preferred). - GitHub private repository security advisories. - The HashiCorp Vault security bulletin (https://discuss.hashicorp.com/tag/security-vault/) - monitored because OpenBao shares a fork point with Vault. rejected_channels: Reports arriving by other mechanisms are referred to the mailing list and are not treated as valid reports. timeline: confirmation_days: 7 patch_days: 90 total_days: 97 detail: >- A 7-day confirmation window from submission determines validity, with an acknowledgement sent quickly by a triaging individual. From confirmation, a maximum 90-day window to patch all affected released versions. High or Critical issues get an announcement on all channels one week before the release ships. embargo: >- In-progress reports are kept confidential on a need-to-know basis. Embargoed patches are not disclosed outside the organisations of Dev WG voting members. Those organisations may run patches on their own infrastructure and may ship patched BINARIES to third parties under strict NDA, but must not ship patch sources or reveal vulnerability details before public disclosure. cve_assignment: A GitHub security advisory with CVE assignment is created on confirmation, which also creates the private fork used for patch development. ai_guidance: url: https://github.com/openbao/openbao/blob/main/AGENTS.md detail: >- OpenBao's AGENTS.md carries explicit rules for AI-assisted security research: do not auto-file reports or write descriptions on a human's behalf, every report must include a detailed functional reproducer (a Go test snippet or a shell script against a dev-mode server), minimise source snippets in favour of reproducers, read the published threat model, and respect the supported-versions policy. Rare enough to be worth naming - most projects have said nothing about this at all. track_record: advisories_published: 30 source: https://api.github.com/repos/openbao/openbao/security-advisories as_of: '2026-08-27' most_recent: - id: GHSA-rh46-vc3j-w2w3 severity: critical published: '2026-08-18' summary: Internal operation dispatching leads to token creation. fixed_in: 2.6.2 - id: GHSA-g892-p242-8g86 severity: low published: '2026-08-18' summary: PKI allowed_ip_sans_cidr not enforced for CSR IP SANs when use_csr_sans=true. fixed_in: 2.6.2 - id: GHSA-34fc-gh42-pj53 severity: critical published: '2026-07-14' summary: Recovery mode vulnerable to token leakage via timing attack. - id: GHSA-59w7-v8rr-pr4p severity: high published: '2026-07-14' summary: Templated policies allow privilege escalation via wildcard characters. note: >- Advisories are published with GHSA IDs and cross-referenced from the CHANGELOG entry for the release that fixed them, so a reader can go from a version to its security content and back. threat_model: published: true url: https://openbao.org/docs/internals/security/ note: Referenced from AGENTS.md as required reading before filing a security report. bug_bounty: present: false note: No HackerOne, Bugcrowd or Intigriti program. Disclosure is unpaid and community-run, which is normal for an LF-hosted project. security_txt: present: false note: >- No /.well-known/security.txt on openbao.org (404 on 2026-08-27, both apex and www). This is the one concrete, cheap gap in an otherwise strong disclosure posture - a five-line security.txt pointing at openbao-security@lists.openssf.org and the CVE policy page would make the program machine-discoverable. Recorded so it can be raised with the project. supply_chain: artifact_signing: CoSign and GPG on binaries; CoSign on container images. sbom: Published per release on the GitHub release page. osps_baseline: https://openbao.org/community/policies/osps-baseline/ security_insights: https://github.com/openbao/openbao/blob/main/.github/security-insights.yml