Skip to main content
v0.4.0
Multi-service debugging and project-level context

Multi-Service AI Debugging Across Frontend and Backend

v0.4.0 is a major step forward for OnCall. This release introduces project-level, multi-service debugging, allowing developers to investigate issues that span frontend, backend, workers, and other processes in a single session.

What’s New

oncall cluster
Debug across multiple services by registering them to a shared local cluster. This enables correlated investigations across services without switching tools or terminals.
oncall init
Group related processes under a single project ID, making it easier to reason about complex systems and share a consistent debugging setup across a team.
Multi-process debugging in one session
Attach multiple running processes and debug them together instead of treating each service in isolation.
Cross-service context expansion
AI investigations now reason over logs and runtime signals from all connected services, not just the process that triggered the investigation.
CLI standardization
The npm package has been renamed from clippy-test to oncall-cli, and the standard CLI command is now oncall.

Why This Matters

Most production bugs are distributed across services.
This release shifts OnCall from process-level debugging to system-level investigation, making it practical for real-world architectures.
v0.3.1
Unlimited log search and deeper AI investigations

Unlimited Log Search and Deeper AI Investigations

v0.3.1 focuses on removing constraints during debugging and making AI investigations more flexible and cost-efficient.

What’s New

Full log search across all available logs
Search without artificial limits on context size, enabling broader exploration before narrowing down a root cause.
Expanded investigation scope
Investigations now go beyond breaking failures and support performance issues, anomalies, and unclear runtime behavior.
More flexible queries
Ask natural questions across logs, code, and runtime signals without rigid formats.
Reduced token usage
Internal optimizations significantly reduce token consumption during investigations.

Why This Matters

Debugging often requires wide exploration before precision.
This release removes friction during that exploration phase while keeping investigations efficient.
v0.3.0
Adaptive, tool-driven AI investigations

Adaptive, Tool-Driven AI Debugging

v0.3.0 upgrades the core investigation engine powering OnCall. AI investigations are no longer fixed scripts and are now adaptive and evidence-driven.

What’s New

Local code and log navigation
During investigations, the AI can navigate local files and logs to ground its reasoning in real runtime evidence.
Remote investigations with local context
Investigations run remotely while securely accessing local logs and code when needed.
Adaptive investigation flows
Investigation paths can branch, loop, and adjust direction based on new evidence, instead of following fixed step sequences.

Why This Matters

Real debugging is non-linear.
This release makes AI behave less like a checklist and more like a thoughtful debugging partner.
v0.2.0
Live AI debugging for running processes

Live AI Debugging for Running Processes

v0.2.0 introduces interactive AI-assisted debugging for running processes, allowing issues to be investigated as they happen.

What’s New

AI-assisted debugging for live processes
Attach OnCall to a running command and investigate runtime behavior in real time.
Split terminal interface
A split view shows live logs alongside AI interaction, removing the need for context switching.
Optional AI reasoning visibility
Developers can inspect how the AI is reasoning during investigations when deeper understanding is needed.

Why This Matters

This release turns debugging into a live conversation with your system instead of a post-mortem exercise.
v0.1.0
Initial release

Interactive AI Debugging for a Single Process

v0.1.0 is the first public release of OnCall, focused on interactive debugging for a single running process.

What’s Included

Interactive AI-assisted debugging
Debug a running command with AI assistance during execution.
Real-time log inspection
Inspect logs as they are produced instead of after failure.
CLI-first workflow
Designed from day one to fit naturally into existing terminal-based developer workflows.
This version was released under the internal working title clippy-test.

Why This Matters

Every system-level debugging tool starts small.
This release laid the foundation for everything that followed.