Release: Nightly -> Main #244
Merged
Conversation
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
- Added retry mechanism with configurable retry count to handle transient failures - Enhanced prompt formatting with clearer template list presentation and validation rules - Improved error handling to retry on failures before falling back to null selection
…tion - Move auto-resume logic to initial state restoration for more reliable reconnection handling - Update agent state handler to only manage UI state, not trigger generation - Add proper phase cancellation tracking when generation is stopped by user
- Fixed model disposal to prevent memory leaks - Improved value updates using pushEditOperations for better undo/redo support
- Process escaped newlines and tabs in commit messages to preserve proper formatting - Apply unescaping in both FileManager and git tool for consistency
- Set isGenerating flag immediately before triggering generation to avoid UI state inconsistencies - Added flag update for both new chat and reconnection scenarios
Fix: general bug fixes, prompt fixes, workflow improvements
- Split monolithic simpleCodeGeneratorAgent into baseAgentBehavior and PhasicAgentBehavior - CodeGeneratorAgent now inherits Agent class (DO) and uses behavior - AgentBehavior relies on AgentInfrastructure (provided by DO) to work, allows to write non-DO coding agents (like cli based) - Phasic for legacy state machine app builder, Agentic for agent driven loop - Agentic would support multiple project types - 'app', 'workflow', 'slides/docs'
…h-github-token-caching
Fix: patch GitHub token caching
…t 1" This reverts commit fc34077.
- Add agent-specific model constraints and filtering - Refactor centralize AI model configuration with credit-based system - Add SimpleCodeGeneration operation - Update model configurations and rate limits - Improve model selector and config UI
AshishKumar4
force-pushed
the
nightly
branch
from
52fa74b to
3f9ccf7
Compare
…ration - Add instance-level static analysis cache to reduce redundant analysis calls - Invalidate cache on sandbox deployment to ensure fresh analysis - Move final phase generation logic from executeFinalization to use executePhaseGeneration - Remove staticAnalysis parameter passing through execution methods - Move LAST_PHASE_PROMPT from PhaseImplementation to PhaseGeneration operation - Add isFinal flag to phase generation to
) * refactor: split agent config by deployment type and reduce resource limits Split AGENT_CONFIG into platform-specific and vanilla deployments based on PLATFORM_MODEL_PROVIDERS environment variable. Add documentation comments explaining config usage. Reduce deepDebugger max tool calling depth from 100 to 50 and conversation compactification threshold from 100k to 50k tokens. Add Google Vertex AI models (GPT OSS 120B and Kimi K2 Thinking) to model catalog. Enhance blueprint system prompt to emphas * refactor: adjust blueprint and implementation agent model configurations Switch blueprint agent from Gemini 3 Pro Preview to OpenAI 5 Mini with reduced max tokens (32k). Update firstPhaseImplementation and phaseImplementation agents to use Gemini 2.5 Pro instead of Gemini 3 Pro Preview. Lower firstPhaseImplementation temperature from 1.0 to 0.2 for more consistent output. Update blueprint system prompt to clarify project seriousness by replacing "toy or educational" with "toy or demo". * refactor: extract shared agent configs and improve config organization
Refactor: improve static analysis caching and finalization phase generation rewrite
…e type exports Move getModelConfigsInfo implementation from SimpleCodeGeneratorAgent to ModelConfigService for better separation of concerns. Export ModelConfigsInfoMessage, AgentDisplayConfig, and ModelConfigsInfo types from websocketTypes for frontend consumption. Remove duplicate AgentDisplayConfig interface from model-config-tabs.tsx. Add blueprint_chunk WebSocket message type. Remove 'advanced' workflow tab and update agent
AshishKumar4
force-pushed
the
nightly
branch
from
b4f0e69 to
7b90c82
Compare
Add null checks after inference calls in blueprint, template selection, phase generation, user conversation processing, project setup, and code fixer operations. Switch from REACT_RENDER_LOOP_PREVENTION to REACT_RENDER_LOOP_PREVENTION_LITE in code debugger and phase generation. Add explicit warning against installing cloudflare:workers/durable-objects dependencies in multiple prompts. Update phase generation to prioritize runtime
Add null checks after inference calls in blueprint, template selection, phase generation, user conversation processing, project setup, and code fixer operations. Switch from REACT_RENDER_LOOP_PREVENTION to REACT_RENDER_LOOP_PREVENTION_LITE in code debugger and phase generation. Add explicit warning against installing cloudflare:workers/durable-objects dependencies in multiple prompts. Update phase generation to prioritize runtime
Upgrade sandbox Docker image from 0.5.2 to 0.5.6. Add instance ID validation with pattern matching and length limits to prevent path traversal. Extract integer argument parsing to dedicated helper function. Improve Result type narrowing with explicit 'error' field checks throughout error and log commands. Convert async storage cleanup to synchronous close() calls. Enhance file lock mechanism in log reset with stale lock detection and retry logic matching
…tions Switch blueprint agent fallback from Gemini 2.5 Flash to OpenAI 5 Mini. Comment out 429 rate limit error throwing in inference core to allow fallback retry logic. Remove null check exceptions in phase generation, phase implementation README generation, code fixer, and user conversation processing - return empty/default values instead of throwing errors. Update phase generation to handle null results with empty phase object
Feat: sandbox and sub-agents stability updates
github-actions
bot
added
the
dependencies
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
3 participants
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
No description provided.