Task graph
Work stays visible.
The task graph shows Completed, Current, and Backlog work with dependencies, blockers, and critical paths.
A work system around the coding agent.
OpenSymphony coordinates task graphs, isolated workspaces, harness execution, review, recovery, and code-aware context in one local-first control plane. The desktop app and TUI make the work, the run, and the knowledge it produces visible to the operator.
Local-first control plane
Dependency-aware task graph
OpenHands + local Codex
Memory + code intelligence
Desktop operator surfaces
OpenSymphony
Operator surfaces
Harness
Execution stays attached to the issue.
OpenHands is the default route. Supported local Codex app-server runs add readiness checks, thread identity, interrupts, and debug resume.
Knowledge
Completed work becomes project memory.
Memory capture and the knowledge graph preserve issue capsules, relationships, source references, and navigable context.
Desktop
The operator can see the system.
The desktop app connects task graph, run detail, changed files, and graph views so long-running work stays inspectable.
Why it matters
The operator sees the work, the run, and the knowledge it leaves behind.
OpenSymphony keeps coordination authoritative while each surface shows the evidence needed to decide what happens next.
01
The task graph organizes the work.
Dependencies, blockers, critical paths, and project state give long-running work a durable operating shape.
02
Runs retain their operational context.
Workspaces, harness sessions, run detail, interrupts, retries, and recovery keep execution inspectable.
03
The system carries knowledge forward.
Memory capture and AST-backed code intelligence connect completed work to the files, symbols, diagnostics, and decisions it informs.
Operator surfaces
See the work at the level where decisions happen.
The desktop app brings task state and project memory into views that can be inspected, filtered, and followed back to their source context.
Task Graph
Dependencies become visible operating context.
Completed, Current, and Backlog panes show blockers, cross-pane relationships, and the critical path through active work.
Knowledge Graph
Completed work remains connected to what the project knows.
The Knowledge Graph groups project memory into navigable areas, concepts, capsules, relationships, and source references.
Workflow
From task graph to code-aware execution.
OpenSymphony moves a work item through a controlled path: dependencies choose the work, an isolated workspace contains it, a harness executes it, and graph-backed surfaces keep the result legible.
01
Select work from the task graph
The orchestrator reads Linear, applies hierarchy and dependency rules, and surfaces Current, Backlog, and Completed work.
02
Create or reuse the workspace
Each issue gets a deterministic workspace with manifests, hooks, prompts, memory, and generated context.
03
Run through the selected harness
OpenHands is the default route, with local Codex app-server support, model readiness, thread reuse, and interrupts when selected.
04
Inspect, review, and recover
The desktop app, TUI, memory graph, and code intelligence expose activity, changed files, symbols, diagnostics, review state, and recovery paths.
Quick start
Start with the repository's Quick Start.
You can follow the instructions without cloning the repository. When you want architecture docs, memory references, or deeper project material, use the Resources page.
Latest updates
Deep dives into the system
July 6, 2026
Graph Surfaces for the Desktop App
Knowledge Graph navigation, a three-phase Task Graph, and the opensymphony app launcher make the desktop control surface easier to inspect and use.
June 21, 2026
OpenSymphony Adds Codex App-Server Harness
OpenSymphony 2.0.0 adds a supported local Codex app-server harness, Codex CLI readiness reporting, and richer Codex run debugging.