EN中文

See the full comparison at docs/comparison.md. Key highlights:

完整对比见 docs/comparison.md。核心要点:

Agent Loop & Execution

核心 Agent 循环与执行

FeatureClaude CodeSeek
Tool loop (read/write/bash/edit)
FIM completions
Think / reflection
LSP references🔶 Slim (refs only)
Background bash + monitor
Session persistence
/goal autoloop✅ (also headless+cron)
Web search❌ (fetch only)
Notebook edit

Planning & Permissions

计划与权限系统

FeatureClaude CodeSeek
Plan mode (analyze→propose→execute)✅ v2
Two-axis permission (pref × workflow)🔶
YOLO / ask-all modes

Subagent & Automation

子代理与自动化

FeatureClaude CodeSeek
Subagent (parallel worktree)
Cron / scheduled wakeup
OS sandbox (Seatbelt/Landlock)

Memory & Skills

记忆与技能系统

FeatureClaude CodeSeek
Three-tier memory (L/M/S)🔶
Skill system (v2)
Soul / cross-project induction

For the full detailed comparison including gap tracking, roadmap priorities, and Reasonix analysis: docs/comparison.md (35KB).

包括差距追踪、roadmap 优先级和 Reasonix 分析的完整对比:docs/comparison.md