Two projects.
One decision layer.

@consensus-tools/wrapper wraps any async function with a consensus gate. Reviewers score the output, strategy decides allow / block / retry / escalate.

const safeDeploy = consensus({
  name: "deploy-to-prod",
  fn: deploy,
  reviewers: [securityReview, complianceReview],
  strategy: { strategy: "threshold", threshold: 0.7 },
});
// result.action => "allow" | "block" | "escalate"

@consensus-tools/guards ships 7 built-in guard types with weighted voting: code_merge, deployment, send_email, publish, support_reply, agent_action, permission_escalation.

const votes = evaluatorVotes({
  action: { type: "code_merge", payload: { files } },
  context: {},
});
const result = finalizeVotes(votes, "code_merge");
// => { decision: "ALLOW", risk_score: 0.34 }

@consensus-tools/workflows is a DAG engine with guard nodes, human-in-the-loop pauses, and cron scheduling.

const workflow = await runner.createWorkflow("PR Review", {
  nodes: [
    { id: "guard", type: "guard", config: { guardType: "code_merge" } },
    { id: "approve", type: "hitl", config: { channel: "slack" } },
    { id: "merge", type: "action", config: { action: "github.merge_pr" } },
  ],
});