Kilo Code Ships JetBrains Plugin 7.1.9-rc.1 With Auto-Recovery for Context-Limit Errors

Kilo Code published a release candidate for its JetBrains IDE plugin, version 7.1.9-rc.1, on September 28. The pre-release patch focuses on reliability fixes for AI agent sessions and several interface refinements, according to the release notes on GitHub.

What changed

The update's most notable fix addresses a recurring failure mode in long agent sessions: when a provider returns a context-limit error, the plugin now automatically compacts the conversation to recover instead of failing the session outright. This targets a common pain point for developers running extended agentic coding workflows where accumulated context exceeds a model's window.

Other fixes in this release include:

• Prompt cache breakpoints are now preserved only for first-party OpenAI providers, preventing the plugin from sending incompatible cache-breakpoint requests to custom or third-party OpenAI-compatible endpoints.
• Scheduled sessions now report their actual wake time rather than showing a generic "busy" status, giving developers clearer visibility into when a paused agent will resume.
• Embedding dimension configuration is now correctly passed to OpenAI-compatible indexing providers, fixing a mismatch that could affect codebase indexing accuracy.

On the feature side, Kilo Code renamed "workflows" to "commands" throughout the JetBrains plugin interface and reworked the associated settings presentation. Agent board messages now render as Markdown rather than plain text, improving readability of formatted agent output. The release also unifies goal-setting with the scheduling and timing tools, letting agents coordinate long-running, multi-step work more reliably across sessions.

Release status

This is a release candidate (rc.1), meaning it is a pre-release build intended for testing ahead of a stable rollout. It was built automatically via GitHub Actions from commit 2ccf28e, three commits ahead of the main branch at time of publication.

What this means

This is an incremental maintenance release, not a new model or major feature launch. The changes reflect a pattern common across AI coding assistants in 2025: as agents run longer and more autonomous sessions, failure handling around context limits, caching, and scheduling becomes the primary engineering burden rather than raw model capability. Automatic conversation compacting on context-limit errors is a practical fix that reduces friction for developers relying on Kilo Code for multi-step, unattended coding tasks inside JetBrains IDEs. Teams already using the JetBrains plugin should expect this rc to promote to stable shortly if no regressions surface during testing.