Remedy Remedy / AI Bug Fixer & Autonomous Software Remediation
From bug report to verified fix.
Your users can report problems. Your software can report problems. Remedy is designed to investigate both—connecting incident evidence to diagnosis, controlled repair, deterministic tests and production verification.
Human + machine incident entryCapability-controlled remediationDeterministic verification
PROBE · GET /checkoutVerified
Repeated 500 after deployment
Runtime, repository and deployment evidence correlated.
01
DetectedAPI probe breached configured assertion
02DiagnosedFailure mapped to deployed source state
03RepairedIsolated candidate passed required gates
04VerifiedProduction probe passed after deployment
ObserveDetectCorrelateDiagnoseRemediateTestReviewDeployVerify
One incident model
A software problem can begin anywhere.
A human report is one route into Remedy. The architecture also covers application exceptions, failed APIs, unhealthy services and connected telemetry, subject to the component availability shown below.
The Remedy ecosystem
Several evidence paths. One repair engine.
The Reporter, SDK and repair foundation exist. The Agent is in development; Probes and connectors have planned or workflow-dependent status. Check the current integration status.
Remedy Reporter
Human-reported problems from web and mobile applications.
Remedy SDK
Bounded application diagnostics and structured evidence.
Remedy Agent
Capability-controlled server and runtime monitoring.
Remedy Probes
API, health and service assertion monitoring.
Remedy Connectors
Repositories, databases, deployment and observability evidence.
→
Remedy Engine
Diagnose → Repair → Test → Verify
Correlates incident evidence, maps exact repository state, prepares an isolated candidate, applies deterministic project gates and follows the approved change into production verification.
EvidenceRoot causeExact SHAIsolated repairTestsReviewVerification
The complete lifecycle
Find it. Diagnose it. Fix it. Prove it.
Remedy does not stop when the patch is written. It follows the defect through testing, review, deployment and deterministic verification.
INCIDENTEVIDENCEDIAGNOSISREPOSITORYISOLATED REPAIRTESTSREVIEWBRANCH / PRDEPLOYMENTVERIFIED
Explore the platform
Start with the part of the system you recognise.
Remedy is written for technical founders and professional engineering teams without requiring either audience to decode the architecture first.
Human entryRemedy ReporterIn-app bug reporting that starts a structured repair process.Application evidenceRemedy SDKBounded diagnostics for web, Node.js and React Native foundations.Runtime · In developmentRemedy AgentServer, service and container evidence with explicit capabilities.Probes · Planned expansionAPI monitoringDeterministic assertions that can begin and verify incidents.Repair engineAI code repairExact source state, isolated change and project-specific gates.OutcomeVerified repairA generated patch is not fixed until deterministic verification passes.OperationsDocker remediationRecover approved containers, then address recurring root causes.Source controlRepository connectionsGitHub, GitLab, Forgejo, Bitbucket and generic Git architecture.EvidenceObservability + remediationUse the evidence needed to understand, repair and verify.
Bounded autonomy
The Agent does not need unrestricted control of your server.
Customers determine which evidence and capabilities Remedy receives. A named service restart is a capability. Unrestricted shell access is not the operating model.
✓Outbound Agent communication by design
✓Explicit service, container and deployment capabilities
✓Read-only database diagnostics as the safe default
✓Human review and customer-governed deployment
// customer-defined capability policy
observe: ["api-health", "service-status"]
operate: ["restart:checkout-api"]
repository: "mapped/exact-sha"
database: "diagnostics:read-only"
deploy: "approval-required"
close_when: "verification-passed"
observe: ["api-health", "service-status"]
operate: ["restart:checkout-api"]
repository: "mapped/exact-sha"
database: "diagnostics:read-only"
deploy: "approval-required"
close_when: "verification-passed"
Clear positioning
AI bug fixing, expanded responsibly.
Is Remedy only a frontend bug-reporting widget?
No. Remedy Reporter is the human-facing entry point. The broader platform includes application evidence, runtime Agent and Probe architecture, repository connections, AI repair and deterministic verification.
Is every component generally available?
No. The site labels the current Reporter and repair foundation separately from Agent, Probe and connector capabilities that are in development or planned.
Does Remedy replace developers or observability platforms?
No. Remedy focuses on evidence-backed diagnosis, repair orchestration, controlled remediation and verification. Engineering judgement and broader telemetry platforms retain their own roles.
Choose your next step
Understand the workflow. Compare the options.
Explore the engineering guide, compare documented product scope or see Remedy’s planned early-access pricing.
LearnHow AI bug fixing worksFrom an observed failure to a testable diagnosis, reviewed change and verified outcome.CompareSix AI bug-fixing toolsRemedy, Tembo, Gitar, Copilot Autofix, Snyk Agent Fix and Cursor Bugbot.Early accessPlans from Free to EnterpriseOne application and server pricing model, with clear component availability.