Announcing the Athena Partner API
View all articlesDavid Henry
Director of Solutions Marketing
Chainguard
David Henry Director of Solutions Marketing
Chainguard now gives Athena shield partners an API to register the mitigations they can deliver and report back on the ones they ship. It's available to Athena partners today through chainctl and REST.
Protected before the patch
Athena is Chainguard’s industry coalition that finds open source vulnerabilities, gives members fixes before public disclosure, and protects them with layered defenses against AI-driven attacks. Last Thursday, Athena disclosed its second wave: 12 silent vulnerabilities across 9 Java libraries, two of them critical authentication bypasses. One let anyone who could reach an Apache Kerby KDC over TLS turn an unsigned JWT into a real administrator ticket. The other let an unauthenticated caller point MockServer's control API at a cloud metadata address from inside a CI runner. Both were fixed upstream years ago. Neither got a CVE, so nobody had a reason to move.
By the time the advisories went public, five shield partners had protection in place. Palo Alto Networks shipped threat-prevention signatures for the network-reachable findings before disclosure, covering every workload behind its firewalls with no restart and no policy change. Akamai and Zafran pushed WAF rules and EDR policies to customers. Cloudsmith turned on blocking policies for Athena members during the embargo, then for everyone. Teams that could patch the same day got a second layer. Teams that couldn't were still protected on day zero.
Shield partners sit in the network path, in front of traffic, or on the endpoint, where they can block an exploit before anyone downstream has rebuilt anything. The second wave was the first time we ran that at disclosure with five partners at once. It worked. It also showed us what to build next.
The gap between a fix existing and a fix being deployed
Athena launched in June, and it has now processed 58,000 findings, 42% critical or high and 86% network reachable. Generating fixes was never the bottleneck. Getting them into running systems was.
Dan wrote in September that the flood is coming and the pipes were already full. Athena ships the patch, advisory, and remediated artifact the same day. That change still has to be scheduled, tested, and rolled out, and plenty of organizations can't do that on an attacker's timeline, or at all. That's why Athena's lifecycle has a step that has nothing to do with patching. Find, fix, shield, surface, disclose. Shield is non-patch protection around a finding while the fix makes its way to deployment, and it's the only step that can move at the speed the attacker does.
In July we described how partners plug in: a Markdown capability manifest we reviewed by hand, a routing filter that matched each finding against every manifest, and partners acknowledging or declining what we sent. That design held through the second wave. Routing embargoed findings only to partners who could mitigate them kept the circle small, and ack/decline tuned what we sent next. Now it needs room to grow. Partners add product lines and new classes of signature, so a manifest should be something they update themselves. And the most useful data in the loop, which findings have a mitigation in place and from whom, should be something members can query while they patch, not something we assemble from email threads and spreadsheets.
The partner program is free, on purpose. The organizations that protect the industry shouldn't have anything in their way, including manual work for every registration and every report. The next wave will have more partners and more findings, and Athena has to scale with them.
So we built the API.
How the Athena Partner API works
The Partner API takes the loop we've been running by hand since July and makes it something partners drive from their own tooling: declare what you can mitigate, get routed the right findings, report what you shipped, and have that coverage show up for members without anyone copying it out of an email.
Partners register the mitigation methods they support from the Athena method catalog: WAF rules, traffic blocks, detection signatures, runtime controls, virtual patches. Methods can be scoped to specific packages or left open to cover everything. This replaces the hand-written manifest with something structured and easy to update when capabilities change.
Chainguard reviews each request. This step stays human. Embargoed findings are the most sensitive data in the program, and a person confirms a partner can act on a method before pre-disclosure details flow. Each method is marked approved or rejected, and partners can check that state at any time.
Approved partners receive findings through the Athena OSV API, limited to the records they can mitigate. It's the same OSV format members already consume, so existing integrations work as they are. Nothing lands in a partner's queue that they can't act on.
Partners report back what they implemented. Chainguard records which findings have a partner mitigation in place and shares that with Athena members.
Same shape as July. Every step a partner touches is now an API call, and the one step that needs judgment is the only one that waits on a person.
Reports show coverage and keep exploit details out
A mitigation report carries one of three statuses per vulnerability: MITIGATED, DECLINED, or FAILED. That's all it carries. No exploit details, no sensitive data, nothing that identifies a partner's customers. It answers one question: whether a mitigation exists for this finding, and that's what makes it safe to share across the coalition.
The other two states are worth as much as the first. DECLINED means a partner looked and chose not to act, which tunes routing the way the old ack/nack did. FAILED means a partner tried and couldn't, which usually means the finding needs a different kind of protection. Together they show where the gaps in the shield are, and that's where we go looking for the next partner.
Partners can create, list, describe, update, and withdraw reports with chainctl athena partner report or through the REST endpoint.
How to get started
You need to be an Athena partner, and your identity needs the Athena partner permissions. Your Chainguard account team can set that up.
Read the Partner API overview, then the capabilities guide and the reporting guide.
Register the mitigation methods you support with
chainctl athena partner capabilitiesor the REST endpoint for your organization. Add package interests, or leave them empty to receive findings for all packages.Check the review state of each method. Chainguard marks each one as approved or rejected.
For each vulnerability you act on, send a mitigation report with
chainctl athena partner reportor the REST endpoint. UseMITIGATED,DECLINED, orFAILED, and keep the note short and free of exploit detail.
The step-by-step is in the Partner Guides. Anything they don't cover, your Chainguard account team can.
What's next
Two waves in, Athena has disclosed 27 silent vulnerabilities. The validated backlog behind them is in the tens of thousands, and the ones still to come won't all be already-fixed bugs in old versions. Every one lands in a world where most deployed software is old, and most teams can't patch on a model's timeline. Upgrading is still the fix. The shield is what covers people until the fix lands.
The second wave showed that five partners can have protection in place at the moment of disclosure. The Partner API is how we get from five partners and twelve findings to every partner who can help and every finding they can act on, without a Chainguard engineer in the middle of each one. Partners register once, receive only what they can mitigate, and report back in a form that's safe to share. Members get a straight answer about their coverage.
Most of the people this protects will never know there was anything to be protected from. That was always the point.
If you sit in front of traffic, run the endpoints, or carry protections into customer environments, there is a role for you in Athena. Apply to join as a shield partner or submitting member.
Share this article
Related articles
productHow Chainguard’s frontier AI harness keeps your code safe
productIntroducing Project Theseus, starting with challoc
productRun GitLab-certified hardened images for GitLab Enterprise Edition with the security of Chainguard
productChainguard endpoints now handle breaking changes
productBuilding Guarded Entrypoint for Chainguard Containers
productIntroducing Chainguard Workspaces