---

Your machine runs 4 different AI agents. Your favorite skill exists as a GitHub repo. Now what? Copy folders into ~/.claude/skills, then ~/.agents/skills, then ~/.gemini/skills, then ~/.copilot/skills… and re-do it after every upstream update. And once every agent has twenty MCP servers wired up, their tool definitions eat your context window alive.

aipx fixes both. One command installs a skill into the shared standard root (~/.agents/skills — read natively by dsh and Codex, linkable by the rest). And the aipx MCP hub fronts ALL your MCP servers with ~4 meta tools — search, call, status, import — so the model sees one server instead of fifty. The bundled skills teach you (and your agents) how to publish for every harness from a single repo.

npx github:zhangliang0115/ai-plugin install <owner>/<repo>

Why

Every agent harness converged on the same skill format — SKILL.md — but not on the same install location:

Agent	Reads skills from
DeepSeek Harness (dsh)	~/.agents/skills/ + <project>/.agents/skills/
Codex CLI	~/.agents/skills/ + <project>/.agents/skills/
Claude Code	~/.claude/skills/ + <project>/.claude/skills/
Gemini CLI	~/.gemini/skills/ + <project>/.gemini/skills/
GitHub Copilot CLI	~/.copilot/skills/ + <project>/.github/skills/
Cursor / OpenCode / OpenClaw	their own roots (full matrix)

Plugins fragment even further: Claude Code wants /plugin marketplace add, dsh wants dsh plugin --profile web add "github:o/r#path:/dsh-plugin", Gemini wants gemini extensions. aipx is the missing common denominator: one installer, one registry, one list, for all of them.

Commands

aipx install owner/repo                          # repo root or skills/ auto-detected
aipx install owner/repo#path:/skills/their-skill # subdirectory (same syntax as dsh)
aipx install https://github.com/owner/repo/tree/v1.2/skills/x   # pinned ref
aipx install ./my-skill                          # local directory
aipx install owner/mcp-server                    # .mcp.json repos add MCP servers too
aipx install owner/repo --project                # project-scoped: .claude/skills,
                                                 # .agents/skills, .github/skills, …
                                                 # committed with the repo for the team

aipx upgrade         # re-install recorded skills from their source (--force semantics)
aipx list            # what's installed, per agent
aipx search deepseek # curated registry; add --github for live GitHub topics
aipx lint skills     # validate SKILL.md quality (frontmatter, triggers, links, nesting)
aipx new my-skill    # scaffold a publish-ready dual-target skill repo
aipx mcp list        # inventory MCP servers across every agent's config
aipx mcp import      # register discovered MCP servers into the aipx hub
aipx mcp serve       # run the hub: one MCP server, ~4 tools, zero context bloat
aipx remove <name>   # uninstall everywhere
aipx doctor          # environment + agent detection report

Example:

$ aipx install JimmyLv/bibigpt-skill#path:/skills/bibi
✔ detected skill with 1 skill(s):
    bibi — Summarize YouTube, Bilibili videos and podcasts…
✔ target roots:
    /Users/you/.agents/skills   (DeepSeek Harness (dsh))
    /Users/you/.claude/skills   (Claude Code)
✔ installed bibi into 2 root(s)

MCP hub — every server, ~4 tools, one context

Every downstream MCP server dumps its full tool catalog into your context. With 20 servers × 10 tools that's tens of thousands of tokens of tool definitions the model must wade through on every turn.

The aipx hub flips it: one MCP server (the hub) fronts all of them and exposes ~4 meta tools. The model searches for a capability, gets the matching tool's inputSchema back, then calls it — loading only what it uses.

aipx mcp import        # pull every MCP server found in your agent configs
aipx mcp serve         # speak MCP over stdio; wire this into any agent:
#   { "mcpServers": { "aipx": { "command": "aipx", "args": ["mcp", "serve"] } } }

Meta tool	Purpose
mcp_search	keyword-search every downstream tool; returns id + description + inputSchema
mcp_call	execute a downstream tool by server/tool id from mcp_search
mcp_status	registered servers, tool counts, health
mcp_refresh	re-scan servers after you add or remove one

Downstream servers are spawned on demand and reused; remote (HTTP) servers and vector search (pluggable index, e.g. a zvec sidecar) are on the roadmap. Docs: MCP hub guide · vector search design.

What's bundled (the toolkit)

This repo is itself a plugin payload — use it three ways:

# 1. Plain skills, every agent:
aipx install zhangliang0115/ai-plugin

# 2. Claude Code marketplace:
#    /plugin marketplace add zhangliang0115/ai-plugin
#    /plugin install ai-plugin-toolkit@ai-plugin

# 3. DeepSeek Harness bundle:
dsh plugin --profile web add "github:zhangliang0115/ai-plugin#path:/dsh-plugin"

Skill	Teaches your agent to
skill-author	write SKILL.md skills that load in every harness — incl. the tier-shadowing and discovery gotchas generic guides miss
skill-portability-audit	audit "works in Claude but not in dsh" failures: collisions, shadowing, trigger quality, per-agent smoke matrix
dsh-plugin-dev	package & publish DeepSeek Harness bundles (cordis.patch.yml, ctx.skills.register, the git-install gotchas)
claude-plugin-dev	publish Claude Code plugins & marketplaces with the dual-target pattern (one repo → every agent)
deepseek-cost-router	route work between deepseek-chat / deepseek-reasoner to cut API cost
deepseek-migration	migrate an agent setup from OpenAI/Anthropic to DeepSeek — caching, tool-calling, cost levers, dsh option

Design principles

• Zero dependencies. One JS file per concern, node:test suite, no supply-chain surface.
• Non-destructive. Installs skip existing targets unless --force; --dry-run previews; removal goes through a manifest.
• One canonical root. ~/.agents/skills is the shared standard (read natively by dsh and Codex) — install writes one copy there and nothing else. No duplicate trees, no drift.
• Context-first MCP. The hub fronts every downstream MCP server with a handful of meta tools; the model searches and calls on demand instead of loading every tool definition into context.

Docs

• Compatibility matrix — every root, every tier
• Install into DeepSeek Harness (dsh) — researched guide: skill roots, tiers, bundle format, gotchas
• Install into Claude Code — marketplaces & plugins
• MCP config sync — 简体中文版：docs/mcp-sync.zh-CN.md
• Publish once, target every agent — the dual-target repo layout
• Troubleshooting — common failures and fixes

Requirements

Node.js ≥ 20 and tar (built into macOS, Linux, Windows 10+). No npm install step — npx github:zhangliang0115/ai-plugin runs straight from the repo. Optional: GITHUB_TOKEN for higher API rate limits.

Roadmap

• v0.1 — install / list / search / remove / doctor
• v0.2 — project-scope installs, aipx new scaffolder, aipx upgrade, lint
• v0.3 — MCP server config sync, registry validation bot + website + install smoke
• v0.4 — MCP hub (mcp import / mcp serve), skills toolkit (6 skills)
• next — vector search for the hub (pluggable index), npm registry publish, registry collections

See ROADMAP.md and CHANGELOG.md.

Contributing

PRs welcome — especially new curated registry entries and community-tier root confirmations. See CONTRIBUTING.md and the plugin submission template.

License

MIT © 2026 zhangliang0115