on 524efea help sign in
7 folders, 24 files
../ agenttest/fix(config): make concurrent config access race-free via copy-on-write hyper/chore: auto-update files notify/fix(server): ignore background errors from unrelated prompts prompt/feat(config): load user-level context files templates/feat(config): load user-level context files testdata/chore(bash/git): improve git commit and PR message standards (#3052) tools/fix(glob): keep file search fast and bounded on large directories .env.sampletest: simplify integration tests, run for a single provider / model accepted_run_test.gofix(server): prevent cancels from affecting future prompts agent.gofix: validate tool call JSON before storing to prevent stuck conversations agent_test.gofix: skip image attachments in history when model doesn't support them (#2818) agent_tool.gochore: gofumpt project agentic_fetch_tool.gofix(glob): keep file search fast and bounded on large directories common_test.gofix(glob): keep file search fast and bounded on large directories convert_test.gofix(agent): prevent session corruption due to malformed image data (#2597) coordinator.gofix(glob): keep file search fast and bounded on large directories coordinator_test.gofeat: add bang mode for direct shell command execution (#3013) dispatch_cancel_test.gofix(server): prevent cancels from affecting future prompts errors.gofeat: support image results from tools (#1549) event.gofix: race in agent.go (#1853) hooked_tool.gofeat: PreToolUse hook (#2598) hooked_tool_test.gofeat: PreToolUse hook (#2598) loop_detection.gofix: detect and stop tool call infinite loops (#2130) (#2214) loop_detection_test.gorefactor: modernize (#2548) prompts.gorefactor(config): introduce ConfigStore and Scope for better config m… (#2395) queued_runid_test.gofix(server): complete queued prompts independently run_complete_test.gofix(server): complete queued prompts independently run_marker.gofix(server): complete background prompts that fail early runid.gofix(noninteractive): crush run reliability in client/server mode usage_fallback.gofix(agent): correct fallback usage accounting usage_fallback_test.gofix(ui): mark estimated context usage