AgentSpaces is security-bearing infrastructure: peers authenticate each other with Ed25519 signatures over self-certifying identities, and the threat model is specified in agentspaces-spec/SPEC.md §11 and the signature inventory in agentspaces-spec/TECH-SPEC.md §4. We take reports against that model seriously.

Please report vulnerabilities privately via GitHub's security advisories:

https://github.com/badmonkeyai/agentspaces/security/advisories/new

Do not open public issues for suspected vulnerabilities, and do not include exploit details in public discussions until a fix is released.

We will acknowledge your report within five business days, keep you informed of progress, credit you in the advisory unless you prefer otherwise, and coordinate the disclosure timeline with you. There is currently no bug bounty program.

Questions can be sent to oss [at] badmonkey.ai

In scope: the Java modules, the Python and TypeScript clients, and anything that lets an attacker violate a guarantee the specs claim — signature or admission bypass, forgery of entries, claims, state transitions or votes, confidentiality failures of group payload encryption, and resource-exhaustion attacks a single peer can mount against the fleet.

Worth knowing before reporting: the specs deliberately document accepted risks and open gaps (agentspaces-spec/TECH-SPEC.md §11, agentspaces-spec/SPEC.md §11) — for example, lying AUCTION bidders and unauthenticated aggregation values. Reports that show a documented accepted risk to be worse than documented are still very welcome.

During 0.x, only the latest minor release receives security fixes.