by provectus
Enables discovery, semantic search, and one‑command installation of AI coding assistant capabilities—including skills, agents, tools, and hooks—directly into a project via a dedicated MCP server and CLI.
Awos Recruitment is a curated registry and MCP server that lets developers locate and install AI‑driven coding capabilities (skills, agents, tools, hooks) using natural‑language queries. It integrates with Claude Code and the AWOS workflow system to automate provisioning of the right components for any development task.
claude mcp add awos-recruitment --transport http --scope project https://recruitment.awos.provectus.pro/mcp
(or add the entry manually to .mcp.json).search_capabilities tool inside Claude Code, optionally filtering by skill, agent, tool, or hook.npx @provectusinc/awos-recruitment@latest skill modern-python-development
npx @provectusinc/awos-recruitment@latest agent testing-expert
npx @provectusinc/awos-recruitment@latest hook skill-update-check
skill status, skill update, skill diff, and skill skip..claude directories..claude/awos-recruitment.lock.json) tracks installed versions and SHA‑256 hashes for reproducibility.AWOS_TELEMETRY=0) to disable anonymous usage reporting.skill-update-check) that notifies developers of outdated or removed capabilities./awos:hire command in AWOS, which combines search, selection, and installation in a single step..mcp.json configuration.Q: Do I need to install anything besides the CLI?
A: Only the npx @provectusinc/awos-recruitment@latest package is required for installation commands. The MCP server is accessed via its URL; no local server binary is needed.
Q: Where are the installed capabilities stored?
A: Skills go to .claude/skills/, agents to .claude/agents/, tools to their respective locations, and hooks to .claude/hooks/ with configuration injected into .claude/settings.json.
Q: How does version tracking work? A: Each installed file’s SHA‑256 hash is recorded in the lock file. The lock also stores the registry commit hash and release date, enabling precise status checks.
Q: Can I keep my local edits when a skill is updated?
A: Yes. Use skill diff --keep <name> to merge registry changes with your edits, then finalize with skill merged <name>.
Q: How is telemetry handled?
A: Anonymous usage counts are sent by default. Set the environment variable AWOS_TELEMETRY=0 to disable this feature.
Q: What if I want to skip a specific version of a skill?
A: Run skill skip <name>; the lock file records the skip, and the skill will be ignored until a newer version appears.

A curated registry of Claude Code skills, agents, tools, and hooks with a smart search MCP server. The Provectus Way of assembling AI capabilities for your project.
Connect the MCP server to Claude Code, search for capabilities using natural language, and install them into your project with a single command.
Works best with AWOS — the AI Workflow Orchestration System for spec-driven development. AWOS drives the SDLC pipeline from spec to implementation; Recruitment provisions the right skills, agents, tools, and hooks for each task. Use the built-in
/awos:hirecommand to fully automate capability provisioning — it searches, selects, and installs everything your project needs in one step.
Claude Code CLI:
claude mcp add awos-recruitment --transport http --scope project https://recruitment.awos.provectus.pro/mcp
We recommend
--scope projectso the MCP config is stored in.mcp.jsonand shared with your team. If you don't need that, change the scope (e.g.--scope user).
Or add to .mcp.json manually:
{
"mcpServers": {
"awos-recruitment": {
"type": "url",
"url": "https://recruitment.awos.provectus.pro/mcp"
}
}
}
Once connected, Claude Code gains access to the search_capabilities tool — search the registry using natural language with an optional type filter (skill, agent, tool, hook).
Install discovered capabilities directly into your project:
npx @provectusinc/awos-recruitment@latest <command> <names...>
| Command | Description |
|---|---|
skill <names...> |
Install skills into .claude/skills/ |
mcp <names...> |
Install MCP server definitions into .mcp.json |
agent <names...> |
Install agents into .claude/agents/ (auto-installs referenced skills) |
hook <names...> |
Install hooks into .claude/hooks/ and inject their configuration into .claude/settings.json |
Examples:
npx @provectusinc/awos-recruitment@latest skill modern-python-development
npx @provectusinc/awos-recruitment@latest mcp context7 playwright
npx @provectusinc/awos-recruitment@latest agent testing-expert
Every installed skill is recorded in .claude/awos-recruitment.lock.json with the sha256 of each file. Commit this file with the skills.
# Compare installed skills with the registry
npx @provectusinc/awos-recruitment@latest skill status
# Update the outdated ones
npx @provectusinc/awos-recruitment@latest skill update
npx @provectusinc/awos-recruitment@latest skill update --dry-run # show the plan only
npx @provectusinc/awos-recruitment@latest skill update react-best-practices
# Replace one skill with the registry version, whatever its state
npx @provectusinc/awos-recruitment@latest skill modern-python-development --force
# Do not offer the current registry version of a skill again
npx @provectusinc/awos-recruitment@latest skill skip react-best-practices
# Show what an update would change (your files against the registry)
npx @provectusinc/awos-recruitment@latest skill diff
# Merge the registry version with your own edits (Claude does the editing)
npx @provectusinc/awos-recruitment@latest skill diff --keep react-best-practices
# ... edit .claude/skills/react-best-practices/ ...
npx @provectusinc/awos-recruitment@latest skill merged react-best-practices
skill diff --keep saves the registry version and a backup of your copy to the project's cache folder, ~/.cache/awos-recruitment/projects/<folder>-<hash>/ ($XDG_CACHE_HOME when set), outside the project, so nothing of it lands in git. After the files in .claude/skills/<name>/ have both the registry's changes and your edits, skill merged records the registry version as the base in the lock file and removes the saved copy. From then on the skill counts as current with your own edits, not as outdated. In Claude Code you only choose "update and keep my edits"; Claude runs these steps.
skill skip records the skipped registry version in the lock file. status shows the skill as skipped by you, and update, the session-start hook and the MCP check stay quiet about it until the registry publishes a newer version. skill update <name> still updates it. When the installed files are the registry's current version, the lock entry also records its registry commit and released date, and status shows them.
skill status prints one of: up to date, outdated, modified locally, removed from registry (with the replacement), no lock entry, not in registry, missing locally. Add --json for scripts.
Skills installed before the lock file existed are matched against the registry's version history (GET /history.json, generated from git at deploy time). A copy that equals a past version shows as outdated with the date of that version and skill update replaces it; a copy that matches no version shows as no lock entry, because it has local changes.
skill update replaces outdated skills and records them in the lock file. A skill you changed locally is not overwritten silently:
update asks: overwrite, skip, or back up to ~/.cache/awos-recruitment/projects/<folder>-<hash>/backups/<name>/<timestamp>/ and overwrite;claude -p), it keeps your copy and says so;--overwrite and --no-overwrite set the answer for every skill.Both commands send anonymous counts (project id from the lock file, skill names, outcome) to the server. Set AWOS_TELEMETRY=0 to turn that off. The server publishes the current hash of every skill and the list of removed ones at GET /index.json.
To be told at session start instead of remembering to run skill status, install the skill-update-check hook. It checks once a day, is silent offline, and prints one line when something is outdated or removed:
npx @provectusinc/awos-recruitment@latest hook skill-update-check
| Document | Description |
|---|---|
| Development Guide | Prerequisites, setup, commands, and project structure |
| Contributing to the Registry | How to add skills, MCP definitions, agents, and hooks |
| Philosophy | Why "Recruitment" and the vision behind the project |
Please log in to share your review and rating for this MCP.
Explore related MCPs that share similar capabilities and solve comparable challenges
by headroomlabs-ai
Compress tool outputs, logs, files, RAG chunks, and conversation history before they reach the LLM, keeping answers identical while saving up to 95% of tokens for JSON payloads.
by modelcontextprotocol
A Model Context Protocol server for Git repository interaction and automation.
by zed-industries
A high‑performance, multiplayer code editor designed for speed and collaboration.
by modelcontextprotocol
Model Context Protocol Servers
by modelcontextprotocol
A Model Context Protocol server that provides time and timezone conversion capabilities.
by cline
An autonomous coding assistant that can create and edit files, execute terminal commands, and interact with a browser directly from your IDE, operating step‑by‑step with explicit user permission.
by upstash
Provides up-to-date, version‑specific library documentation and code examples directly inside LLM prompts, eliminating outdated information and hallucinated APIs.
by daytonaio
Provides a secure, elastic infrastructure that creates isolated sandboxes for running AI‑generated code with sub‑90 ms startup, unlimited persistence, and OCI/Docker compatibility.
by continuedev
Enables faster shipping of code by integrating continuous AI agents across IDEs, terminals, and CI pipelines, offering chat, edit, autocomplete, and customizable agent workflows.