Skip to content

Latest commit

 

History

History
83 lines (55 loc) · 3.62 KB

File metadata and controls

83 lines (55 loc) · 3.62 KB

Roadmap

Ore Code is in pre-release development. This roadmap describes intended direction, not a promise of dates or final scope.

Current Status

The project is being prepared for source-level public GitHub hosting.

Completed foundation work includes:

  • Tauri desktop shell for macOS and Windows-oriented workflows.
  • DeepSeek-first agent runtime with plan, agent, and full-access modes.
  • Structured tool execution, approval flows, skills, MCP integration, automations, project indexing, and Git diff review.
  • Repository entry docs, contribution/security/support notes, known limitations, and a simple CI workflow.

The project is licensed under MIT. Binary release readiness is still separate from source availability.

0.1.0 Source Snapshot

The first public source snapshot should stay small and credible:

  • Documented install, development, troubleshooting, support, privacy, and security paths.
  • Simple CI for TypeScript/package checks.
  • Clear known limitations for MCP, sandboxing, Windows packaging, performance, and public launch caveats.
  • No binary release claims until installers are built and smoke-tested on their target OS.

No public announcement should imply production readiness.

0.1.x Stabilization

After the first public pre-release, prioritize fixes that make day-to-day use dependable:

  • Startup and long-session rendering performance.
  • Windows path, process, shell, line-ending, and packaging issues.
  • MCP reconnect, configuration, and failure recovery.
  • Skills discovery, editing, installation location, and cross-platform behavior.
  • Approval, sandbox, and tool-output UX polish.
  • Documentation updates driven by real user setup problems.

Breaking runtime event, tool schema, persisted data, and settings changes should remain rare in this phase.

0.2 Workflow Depth

The next larger feature wave should improve coding quality and context selection:

  • Persistent project index and incremental indexing status.
  • Symbol graph, call relationship, and impact analysis.
  • Better codebase retrieval and working-set recall.
  • Stronger subagent orchestration and role-specific task routing.
  • Deeper GitHub, PR, and CI workflow integration.
  • Optional sandbox executor with quiet defaults and clear boundary prompts.

These areas should land behind focused tests and compatibility notes rather than as broad UI rewrites.

Future Ideas

Potential later work includes:

  • More provider integrations after DeepSeek-first workflows are stable.
  • Richer MCP marketplace/configuration flows.
  • Team-oriented review and shared policy controls.
  • Linux packaging after macOS and Windows are validated.
  • More complete localization once the UI message boundaries are stable.

Non-Goals for Early Releases

Early public releases should not try to solve every IDE or cloud-agent workflow:

  • No silent system-level dependency installation.
  • No unsupported promise of full sandbox isolation.
  • No broad plugin marketplace commitments before MCP and skills are stable.
  • No Linux desktop release target until packaging and smoke tests exist.
  • No binary release claims until installers are built and smoke-tested on their target OS.

How to Influence Priority

Use GitHub issues after the repository is public:

  • Bug reports should include OS, version, reproduction steps, logs with secrets removed, and whether the issue affects macOS, Windows, or both.
  • Feature requests should describe the workflow, current workaround, and expected benefit.
  • Security-sensitive reports must follow SECURITY.md.

Use Contributing, Support, and Security when turning feedback into follow-up work.