skills
/
deerflow2
3
0
Fork 0
Code Issues Pull Requests Packages Projects Releases Wiki Activity
deerflow2/.planning/ROADMAP.md

4.1 KiB
Raw Permalink Blame History

Roadmap: DeerFlow Frontend Merge Recovery

Created: 2026-04-07 Source: .planning/PROJECT.md + .planning/REQUIREMENTS.md

Phase 1: Conflict Inventory and Decision Matrix

Goal: Build an auditable conflict inventory covering merge-overwritten regions, Titan-overlap regions, and keep/replace candidates.

Covers: MERGE-01, MERGE-03

  • Produce file-level conflict matrix with risk levels
  • Mark visual-only, logic-only, and mixed-change files
  • Identify Titan-linked hotspots and expected behavior per hotspot

Phase 2: Thread and Skills Logic Reconciliation

Goal: Reconcile thread bootstrap/routing and skills API logic so merged behavior is explicit and stable.

Covers: MERGE-02, LOGIC-03, LOGIC-04

  • Reconcile thread_id/isnew/xclaw_used behavior in chat flow
  • Decide and implement skills bootstrap contract direction
  • Verify no duplicate or dead logic paths remain

Phase 3: Legacy Visual Alignment Pass

Goal: Align workspace visual presentation to legacy baseline without regressing logic.

Covers: UI-01, UI-02, UI-03

  • Apply visual alignment in layout and component layers
  • Keep behavioral code intact while adjusting style semantics
  • Validate visual consistency across core workspace views

Phase 4: Iframe + Markdown New-System Stabilization

Goal: Keep and harden new-system capabilities for iframe communication and markdown export.

Covers: LOGIC-01, LOGIC-02

  • Stabilize parent/child messaging and selected-skill event flows
  • Stabilize markdown conversion and download triggers
  • Validate artifact integration points

Phase 5: Test Hardening and Commit Hygiene

Goal: Lock recovery with regression tests and clean commit structure.

Covers: TEST-01, TEST-02, TEST-03

  • Finalize and run E2E suite for target scenarios
  • Split commits into style / logic / tests concern buckets
  • Attach reviewer-oriented verification notes for high-risk files

Phase 6: 在输入框输入@时，可引用已生成文件和已上传附件

Goal: 在当前线程聊天输入框中实现 @ 文件引用（artifacts + uploads），并通过 additional_kwargs.files 稳定提交且具备回归测试。 Requirements: ATREF-01, ATREF-02, ATREF-03, ATREF-04 Depends on: Phase 5 Plans: 4 executable plans + 1 archived revision record

Plans:

  • 06-01-PLAN.md — 锁定引用提交契约与软失败链路（additional_kwargs.files）
  • 06-02-PLAN.md — 实现 @ 候选 dropdown、chip 交互与上限控制
  • 06-03-PLAN.md — 补齐自动化验证并产出 style/logic/tests/docs 提交分组计划
  • 06-04-ARCHIVED.md — 修订归档：原 gap-closure 计划与锁定决策 D-08（上限 10）冲突，保留追踪但不再执行
  • 06-05-PLAN.md — 关闭 verification 缺口：恢复 10 个上限/类型去歧义，并稳定 DF-INPUT-008/009 回归

Phase 7: 发送时拼接附件与Skill优先提示词并在消息区过滤

Goal: 发送消息时拼接附件/Skill优先提示词，同时消息区仅展示用户原文。 Requirements: P7-01, P7-02, P7-03, P7-04 Depends on: Phase 6 Plans: 2/2 plans complete

Plans:

  • 07-01-PLAN.md — 提交态增强文本组装 + 三入口统一透传 + 显示态/提交态分离回归
  • 07-02-PLAN.md — gap closure：修复 ContextMenu 自动引用、提示前缀唯一化、Skill 使用 id 拼接

Phase 8: 现在系统中有非常多写死的颜色值比如bg-[#00000],text-[#000000]，我想把这些颜色值都提升到浅色模式和深色模式里面

Goal: 将 workspace 核心页面/组件中的硬编码颜色迁移为 light/dark 主题 token，并建立防回归扫描守卫。 Requirements: P8-01, P8-02, P8-03, P8-04 Depends on: Phase 7 Plans: 4 plans

Plans:

  • 08-01-PLAN.md — 建立颜色 token 注册表与扫描守卫基础能力
  • 08-02-PLAN.md — 迁移 chat/input/workspace 关键页面组件的硬编码颜色
  • 08-03-PLAN.md — 迁移 artifact 关键组件的硬编码颜色与局部样式变量
  • 08-04-PLAN.md — 建立回归验证闭环并固化防回归检查

Milestone status: complete Next command: /gsd-new-milestone