目标与恢复 · OpenProgram Docs

OpenProgram Docs

Get started Install Capabilities Interfaces Models Integrations Server & Ops Reference Design

Architecture6

Design guide Framework overview Repository structure Implementation index Framework comparison Feature matrix

Agents and workflows83

Execution13

Runtime Codex capability audit Call flow Execution control Action lifecycle Worktrees Recursion diagram Recursion guard Async tasks Nested LLM streaming Task controllability Dispatcher structure Next-step decisions

Sessions10

Reading guide Data model Operations Context Context creation Working directories Management comparison History distillation Storage consistency Title generation

DAG and collaboration14

Design overview LLM node content Agent communication Collaboration tools Collaboration comparison Branch collaboration Layout specification Program nodes Agent DAG ownership Rendering specification Branch naming Naming diagram Edge naming Caller identity

Goals and recovery8

Goals and recovery Resource quotas Generation recovery Tool result storage Stream recovery Progress recovery Web reliability Frontend processes

Configuration and operations9

Agent settings Core settings Tool settings Capability settings Agent implementation plan Chat search Toolsets Input requests Files and recovery

Programs and workflows13

Functions Program model Function calling Fixed-order calls Application runtime Program structure Workflow overview Program responsibilities Workflow authoring Program interface Workflow control Execution unification Self-updates

Reports4

Report workflows Weekly reports Group reports Report delivery

Events and scheduling12

Proactive execution Event layer Event flow diagram Events and state Rule execution Scenario overview Rule templates Invariants Event reference Signal routing diagram Signal routing Memory interface

Context and memory23

Context9

Reading guide Overview Composition Compaction Compaction diagram Occupancy Accounting rules Memory introspection Design comparison

Memory14

Reading guide Architecture overview Architecture diagrams Entity memory proposal Virtual memory Authority and pairing Authority comparison Design adoption Write comparison Editor and history Settings proposal Speaker identity Write cursor diagram Write cursor

Interfaces63

Foundations15

Web UI App icon UI invariants UI surfaces State stores Themes Window state Center canvas layout Tab identity Window lifecycle Style organization Interaction feedback Page loading performance Status indicators Session status

Chat and composer18

Attachment options Attachment delivery Chat turn layout Tool group summary Transcript following Composer interaction Attachment paths Responsive controls Tool profile menu Speed control GUI Agent Agent control overlay Browser task control Queue and steering Session naming Slash commands Turn occupancy WebSocket lifecycle

Browser and tabs8

Browser extensions Built-in browser Session resources Tab placement Remote access diagram Remote access Browser home Native page bounds

Settings and catalog4

Avatars Program explorer Source categories Settings navigation

Workspace and sidebar8

Integrated terminal Project order Project location File icons Project workspace File tree Right sidebar typography Sidebar sizing

CLI and TUI10

CLI and TUI Configuration safety Function execution Naming conventions Port configuration Configuration interface Single-port architecture Command comparison Slash commands Transcript rendering

Models and connections41

Requests and models11

Providers Model catalog Request construction OpenAI compatibility Fast tier Thinking effort Structured output Network proxy Metadata diagnostics Bailian catalog Usage metering

Authentication7

Credential resolution Credential connections File credentials Credential status Credential validation Account rotation Credential storage

Reliability7

Errors and timeouts Retry decisions Error propagation Fault tolerance Recording and replay Replay diagram Reasoning replay

Extensions and integrations13

Integrations Direct calls, Agent classes and automatic DAG context Editor integration Extension management Harness standard MCP tools MCP server Skills and plugins Web specification Web automation Extension controls Future work Reference comparison

Channels3

Channels Requirements Architecture

Security and engineering16

Security and permissions4

Authority and sandbox System access SSRF protection Dependency security

Distribution4

Packaging and releases Automatic updates Developer ID Windows support

Testing and errors3

Test system Windows CI Error handling

Documentation and website5

Documentation site Site discovery Agent-readable docs IndexNow discovery Community discovery

Supporting material23

Prototypes6

Density prototype Composer prototype Workspace prototype Sidebar prototype Sidebar Luma prototype Live layout

Implementation records14

Interface synchronization Repository migration IndexNow delivery Documentation gaps Engineering backlog UI unification Resource preview Distribution plan Web implementation Implementation map Credential migration MCP implementation Cache control Proactive implementation

Research3

Evaluation Policy replay Threat model

Uncategorized12

Overview1

Repository integration

Providers2

Anthropic API authentication Local model providers

Runtime2

Tool name resolution Unified Agent execution

Web UI7

Chat quotes and message editing Conversation lifecycle and responsiveness Effort picker Git folder pills History rendering and conversation links Hover tips in the composer Icon system

Design › Agents and workflows › Goals and recovery › Goals and recovery

Last updated · 2026-10-10

PreviousCaller identity NextResource quotas