Add kilocode_change markers to the Bedrock Claude condition in
transform.ts to ease upstream merges. Move the bedrock-specific
tests to test/kilocode/ so the shared transform.test.ts stays
closer to upstream and reduces merge conflicts.
Implement end-to-end remote session tracking so the Kilo server knows
which sessions are opened and focused in the VS Code extension.
- Add RemoteStatusService that tracks open/focused sessions and pushes
state to kilo serve via the connection service
- Extend connection-service with viewed-sessions endpoint and auto-flush
on remote enable/disable and CLI reconnect
- Redesign remote settings UI in ExperimentalTab with status indicator
and descriptive copy
- Add SDK schema for kilo-sessions.remote-status-changed event
- Gate session reporting on remote-enabled setting; reset state on
CLI disconnect
- Remove vestigial remoteToggled message type
- Add comprehensive unit tests for RemoteStatusService
- Add i18n keys for new remote settings UI across all locales
Add kilocode_change markers to all Kilo-specific changes in shared
opencode files to pass the check-opencode-annotations CI check.
Also bring in the workflow and script from main.
Collect warnings during config loading instead of crashing when config
files have invalid JSON, schema errors, or frontmatter issues. Warnings
are exposed via a new GET /config/warnings pull-based endpoint.
- Add Config.Warning schema and warnings collection in stateInit()
- Wrap all loadFile/load calls with try/catch to collect warnings
- Thread warnings through loadCommand/loadAgent/loadMode/invalid
- Wrap well-known remote config block to prevent network/parse crashes
- Add GET /config/warnings server route with OpenAPI spec
- Regenerate SDK with config.warnings() method
- TUI: fetch warnings in bootstrap non-blocking phase, show as toast
- VS Code: fetch warnings on SSE connect, show as native notifications
- Add 5 new tests for warning collection (JSON, schema, markdown)
The skill loaded but the agent still only searched the repo root when
asked 'where is /review-pr?'. Add an explicit procedural subsection
that tells the agent to search ~/.config/kilo/, ~/.kilo/, ~/.kilocode/,
and ~/.opencode/ with explicit path arguments before falling back to
project directories. Add a regression test verifying the guidance is
present in the builtin skill output.
Docker containers spawned by MCP servers configured with `docker run`
were not cleaned up after being stopped, causing exited containers to
accumulate and fill up Docker storage over time.
Closes#8103
Users launching a follow-up implementation session now get an explicit plan-file path and instruction to read it first, so execution stays aligned with the canonical plan even when copied plan text drifts.
Add 4 tests that reproduce the crash when agent or command markdown
files have valid YAML but invalid Zod schema values (e.g. mode: banana).
These currently pass against the fix that will follow.
Previously, provider HTTP requests had no application-level timeout
unless explicitly configured. Combined with Bun's fetch timeout being
disabled, requests could hang indefinitely when the network was down,
waiting for the OS TCP stack to give up.
Add REQUEST_TIMEOUT_MS (120s) as the default timeout applied via
AbortSignal.timeout() in the provider fetch wrapper. Recognize
TimeoutError as a network disconnection so it enters the offline
prompt flow.
Treat wrapped provider connection errors as offline waits so sleep and network-loss sessions prompt to resume instead of falling through generic retries.
Johnny Amancio