Scans the skills visible to the current session and ranks them by relevance to the recent conversation.
Install
# from GitHub (first run asks for allowBuilds approval — follow the hint, retry)
dsh plugin --profile web add github:hellosky983/dsh-skillradar
Any plugin you install runs third-party code with your own permissions — it can read your files, use your credentials, and reach the network, and tool approvals don’t sandbox it. GitHub-sourced plugins also run build scripts at install time — pnpm blocks those until you allow them, so an install can stop with ERR_PNPM_GIT_DEP_PREPARE_NOT_ALLOWED or ERR_PNPM_IGNORED_BUILDS; dsh prints the exact key to add under allowBuilds in your profile’s pnpm-workspace.yaml, and the install works on the next run. Allowing a build is a trust decision: only install sources you trust, and pin a commit (github:owner/repo#sha).
README
A DeepSeek Harness plugin that scans every skill visible to the current session, scores each against the recent conversation text (English + Chinese token overlap), and returns a ranked recommendation of which skill to load next.
Overview
DSH sessions expose many skills, and neither the model nor the user always knows which one applies. This plugin answers "what should I load right now?" with a concrete, ranked answer. It suits agents and users who work across many skill domains and want a fast, deterministic relevance signal.
Compatibility
- DSH version:
0.1.0-rc.6 - Mainline: verified against
deepseek-harnessmainline snapshots of 2026-08-14 - Last verified: 2026-08-14
Install / Uninstall
Install (from GitHub):
dsh plugin add github:hellosky983/dsh-skillradar
Or clone and install locally:
git clone https://github.com/hellosky983/dsh-skillradar.git
cd dsh-skillradar
dsh plugin add .
Upgrade: re-run the install command after git pull.
Disable temporarily: remove the plugin row from your profile composition, or run:
dsh plugin remove dsh-skillradar
Uninstall: remove the dsh-skillradar dependency and its bundle entry from the profile package.json, then pnpm install.
Quick start
After install and restart, tell the agent:
Scan the current session and tell me which skill fits this task.
or invoke the tool directly:
skill_radar # with no arguments, scans the current session
Example output:
Skill Radar — 16 skills visible
100% github-upload [github, 仓库, readme, 上传]
85% cordis-plugin-development [client, host, cordis, run]
Configuration
| Option | Type | Default | Description |
|---|---|---|---|
session_id |
string | current session | Session to scan instead of the current one |
limit |
integer | 15 | Maximum number of skills returned |
No environment variables, no secrets, no external configuration files.
Permissions & data
- Reads: the current session's recent message text (last 8000 chars) and the skill catalog via
ctx.skills/ctx.sessionQuery. - Writes: nothing. The tool only returns a scored list.
- Network: none. All scoring is local text analysis.
- Credentials: none.
Troubleshooting
- Tool not visible after install: restart dsh, or check that the bundle row appears in
dsh --dump-config. - "cannot get property tools without inject": the plugin requires
inject = ['tools']; reinstall from the latest commit which declares it. - Logs: dsh profile logs under
~/.dsh/profiles/<name>/; setDEBUG=cordis*for component tracing. - Rollback: revert to the previous plugin version by reinstalling the earlier git commit.
Development
git clone https://github.com/hellosky983/dsh-skillradar.git
cd dsh-skillradar
# smoke test with a mock ctx:
node --input-type=module -e "import { apply } from './index.js'; apply({get:()=>undefined, tools:{register:()=>{}}}); console.log('ok')"
The client/ directory contains the optional interactive radar panel (dynamic Cordis plugin form) for the web UI.
Contributions: open issues and pull requests on GitHub. Keep changes dependency-free (scoring is self-contained). Report bugs with a minimal reproduction case.
License & security
MIT — see LICENSE. No secrets are shipped; report security issues privately via GitHub issue (repo is public) or email to the repository owner.
Links
More in this category
vectorize-io/hindsight#coding-agents★ 60916
Hindsight, agent memory that learns: long-term project memory with auto recall and retain, knowledge pages, deep reflection, and per-repo memory banks.
volcengine/OpenViking#examples/dsh-memory-plugin★ 42209
OpenViking memory and context bundle for DeepSeek Harness: pre-step auto-recall and profile injection, session capture, `viking://` URI guarding, and recall/write memory tools backed by an OpenViking server.
MemTensor/MemOS#memos-local-plugin★ 12392
Local-first memory plugin for DeepSeek Harness: layered L1/L2/L3 memory with automatic capture, manual memos_* retrieval tools, and a built-in memory viewer.
agentscope-ai/ReMe#dsh★ 3799
Connects DeepSeek Harness to ReMe's local-first, self-evolving personal knowledge base: automatically captures completed main-agent conversations as user-owned Markdown memory, searches conversations and source material through reme_search with BM25, optional embeddings, and wikilink expansion, and schedules daily memory consolidation.
zilliztech/memsearch#MemSearch★ 2873
Shared Markdown memory for DSH and other coding agents, with automatic capture, pre-step context injection, searchable recall, and memory-to-skill self-evolution through a review panel.
vshulcz/deja-vu#extensions/dsh★ 1544
Cross-session memory and context recovery across forty-five other coding agents (45+) on this machine, reading session files they already wrote before install (Claude Code, Codex, Cursor, Windsurf, Trae, Antigravity, VS Code Copilot Chat, opencode, OpenClaw, Hermes, Kimi, Qoder, Cline, Zed, Magpie and more): six tools (deja_recall, deja_session, deja_blame, deja_fix, deja_how, deja_remember), a /deja command, and optional automatic recall injected into runtime context. Local BM25 CLI/MCP index, no LLM, no embeddings, no network (dsh plugin --profile web add dsh-deja).
Community comments
Comments are public GitHub Discussions. Loading them connects to GitHub and Giscus; a GitHub account is required to post.