YYTbit/dsh-plugin-claude-bridge
Bridge Claude Code memory, skills, and config into DeepSeek Harness
dsh-plugin-claude-bridge bridges Claude Code's memory, skills, and global configuration into DeepSeek Harness — zero migration, full compatibility. It reads Claude Code's standard file locations directly (no file copying or symlinks) and injects them as dynamic context into DSH conversations. Three integration points: (1) Memory — reads all memory markdown files from ~/.claude/projects/<project>/memory/*.md, sorts them by type priority (feedback > project > reference > user), and injects them as a dynamic system prompt context section re-read on each request so new memories take effect immediately; (2) Skills — discovers all skills from ~/.claude/skills/<name>/SKILL.md and injects their names and descriptions into the system prompt as a catalog; (3) Global instructions — injects ~/.claude/CLAUDE.md as global system prompt instructions. All three are individually toggleable (enableMemory / enableSkills / enableGlobalInstructions). Configurable limits: maxMemoryBytes (default 8192), maxSkills (default 30). Extra skill directories can be added via extraSkillDirs. Use it to bring your accumulated Claude Code knowledge base into DSH without any migration work.
Install
dsh plugin --profile web add dsh-plugin-claude-bridgenpm package dsh-plugin-claude-bridge 0.1.1 (registry-verified 2026-08-23). Install: dsh plugin --profile your-profile add dsh-plugin-claude-bridge (replace 'your-profile' with your actual profile name, e.g. 'web'). Works out of the box with zero configuration — all options are optional and have sensible defaults. Optional config in the profile's cordis config: claudeHome (~/.claude), enableMemory (true), maxMemoryBytes (8192), enableSkills (true), maxSkills (30), enableGlobalInstructions (true), extraSkillDirs (additional skill directories).
Compatibility
DeepSeek Harness any profile. Reads Claude Code's standard file locations directly — no migration scripts, no file copying, no symlinks. Reads ~/.claude/projects/<project>/memory/*.md (injects memories as dynamic system prompt context), ~/.claude/skills/<name>/SKILL.md (adds skills to the available catalog), ~/.claude/CLAUDE.md (injects global instructions into system prompt). Memory context is re-read on each request so new memories take effect immediately. Skill catalog entries (names + descriptions) are injected into the system prompt. Zero migration: your existing Claude Code setup works as-is.
Details
- Repo: YYTbit/dsh-plugin-claude-bridge
- Category: Coding & Development
- Stars: 9
- Version: npm package dsh-plugin-claude-bridge 0.1.1 (registry-verified 2026-08-23)
- Last push: 2026-08-14
- First seen: 2026-08-13
Recent updates
The current English README (README.md, primary English) documents: three integration points (memory injection, skill catalog, global instructions), zero-config defaults with optional YAML overrides, memory type priority order, re-read-on-request behavior, install command, full configuration options.
FAQ
- Do I need to copy or migrate any files from Claude Code?
- No — the plugin reads Claude Code's standard file locations directly. Your existing ~/.claude/ directory works as-is; no migration scripts, file copying, or symlinks needed.
- Do new Claude Code memories appear in DSH immediately?
- Yes — memory files are re-read on each request, so any new memories you add to ~/.claude/projects/<project>/memory/ take effect in the next DSH conversation turn.
- Can I disable memory injection but keep skill catalog?
- Yes — each integration point (enableMemory, enableSkills, enableGlobalInstructions) is independently toggleable in the plugin config. All default to true; set any to false to disable that injection.
Alternatives
omdsh-dev/dsh-plugin-skills · shuguang1994/project-blueprint · omdsh-dev/dsh-advisor