Enables agents to search for, audit, and install AI agent skills and MCP servers related to PostgreSQL, with security grades and quality scores.
Click on "Install Server".
Wait a few minutes for the server to deploy. Once ready, it will show a "Started" state.
In the chat, type @ followed by the MCP server name and your instructions, e.g., "@AgentSkillsHub MCP Server search for safe mcp servers to manage postgres"
That's it! The server will respond to your query, and you can continue using it as needed.
Here is a step-by-step guide with screenshots.
๐ Website ยท ๐ Browse Skills ยท ๐ก๏ธ Security Report ยท ๐ข Enterprise
Search, audit, and install open-source AI agent skills & MCP servers from inside your agent โ Claude Code, Cursor, Cline, Cherry Studio, or any MCP client. Every result is security-graded and quality-scored by AgentSkillsHub, a directory of 100K+ skills. The trust signal comes before you install.
// add to your MCP client config
{
"mcpServers": {
"agentskillshub": {
"command": "npx",
"args": ["-y", "@agentskillshub/mcp"]
}
}
}That's it โ no API key, no signup. The server downloads a static catalog index once (~1.7 MB, cached locally) and does all searching locally, so it's fast, works offline after the first run, and puts zero load on the Hub backend.
Discovering a skill is easy. Knowing whether it's safe to run against your credentials is not. This server puts search + a trust check right in the agent's tool loop:
need a capability โ search_skills โ audit_skill โ get_skill_install โ installYour agent sees the security grade and estimated token cost of a skill before it picks one โ signals other directories don't give it.
Related MCP server: skillet
Tool | What it does |
| Find skills by natural-language query + filters ( |
| Free basic trust check for an |
| Install commands for a runtime + a "check before you install" safety line. Returns instructions; it does not run anything. |
User: find me a safe way to query Postgres from Claude Code
The agent calls search_skills({ query: "query postgres", category: "mcp-server", max_security_risk: "safe" }) and gets back graded results:
call518/MCP-PostgreSQL-Ops 150โ
๐ข SAFE quality 75/100
sgaunet/postgresql-mcp 6โ
๐ก CAUTION ~17.2k tokthen audit_skill / get_skill_install before it installs anything.
๐ข SAFE ยท ๐ก CAUTION ยท ๐ด UNSAFE ยท โ REJECT ยท โช UNAUDITED
โช UNAUDITED is not "probably fine" โ it means no one has audited it. The search_skills filter max_security_risk excludes un-audited skills, never silently treats them as safe.
We security-graded the whole catalog and wrote up what we found: We security-graded 117,854 AI agent skills.
Free (this server): search_skills ยท audit_skill (basic) ยท get_skill_install, for any catalogued skill.
Pro / Enterprise: 5-dimension deep audit (code ยท credentials ยท vendor ยท supply-chain ยท operational), any GitHub URL (incl. <5โ / private), CI/batch auditing, compliance evidence โ https://agentskillshub.top/enterprise/
Var | Default |
|
|
|
|
CLI: @agentskillshub/cli โ the same search/audit/install from your terminal (npx @agentskillshub/cli search "โฆ").
MIT ยฉ AgentSkillsHub
Free basic trust check for a skill: security grade, plain-English verdict, and quality score. Run this BEFORE installing/trusting a skill. Deep 5-dimension audit and auditing any GitHub URL (incl. <5โ / private) are Pro features.
| Name | Required | Description | Default |
|---|---|---|---|
| depth | No | Only 'basic' is available for free; 'deep' points you to Pro. | |
| target | Yes | Skill as 'owner/repo' (e.g. 'modelcontextprotocol/servers'). |
Does the description disclose side effects, auth requirements, rate limits, or destructive behavior?
No annotations provided, so description carries full burden. It discloses that only 'basic' depth is free, deep is Pro, and mentions outputs. Lacks details on rate limits or auth, but is sufficient for a simple check.
Agents need to know what a tool does to the world before calling it. Descriptions should go beyond structured annotations to explain consequences.
Is the description appropriately sized, front-loaded, and free of redundancy?
Two sentences, front-loaded with purpose, followed by limitations. No fluff.
Shorter descriptions cost fewer tokens and are easier for agents to parse. Every sentence should earn its place.
Given the tool's complexity, does the description cover enough for an agent to succeed on first attempt?
No output schema, but description lists three output types. Covers limitations (Pro features, free vs paid). Could mention error handling or format of verdict, but adequate for a basic tool.
Complex tools with many parameters or behaviors need more documentation. Simple tools need less. This dimension scales expectations accordingly.
Does the description clarify parameter syntax, constraints, interactions, or defaults beyond what the schema provides?
100% schema coverage sets baseline at 3. Description adds value by clarifying that 'depth' enum value 'deep' requires Pro, and that 'target' format is 'owner/repo'. This goes beyond schema descriptions.
Input schemas describe structure but not intent. Descriptions should explain non-obvious parameter relationships and valid value ranges.
Does the description clearly state what the tool does and how it differs from similar tools?
The description clearly states it performs a 'free basic trust check for a skill' and lists specific outputs (security grade, verdict, quality score), distinguishing it from sibling tools like get_skill_install and search_skills.
Agents choose between tools based on descriptions. A clear purpose with a specific verb and resource helps agents select the right tool.
Does the description explain when to use this tool, when not to, or what alternatives exist?
Explicitly instructs 'Run this BEFORE installing/trusting a skill' and notes when to use alternatives (deep audit and GitHub URL auditing are Pro features).
Agents often have multiple tools that could apply. Explicit usage guidance like "use X instead of Y when Z" prevents misuse.
Get install commands for a skill plus a 'check before you install' safety line. Does NOT run anything โ returns what to install, how, and what to verify first. The agent/user executes.
| Name | Required | Description | Default |
|---|---|---|---|
| runtime | No | Target runtime โ decides which install command is recommended. | |
| repo_full_name | Yes | Skill as 'owner/repo'. |
Does the description disclose side effects, auth requirements, rate limits, or destructive behavior?
With no annotations, the description carries full burden. It clearly states non-execution and the inclusion of a safety verification step. Additional details on response format or side effects would improve, but current description is adequate.
Agents need to know what a tool does to the world before calling it. Descriptions should go beyond structured annotations to explain consequences.
Is the description appropriately sized, front-loaded, and free of redundancy?
Two sentences are concise and front-load the key action. No wasted words, though the second sentence could be more integrated. Marginally below 5 due to slight redundancy.
Shorter descriptions cost fewer tokens and are easier for agents to parse. Every sentence should earn its place.
Given the tool's complexity, does the description cover enough for an agent to succeed on first attempt?
Given no output schema, the description should explain return value structure. It broadly states 'what to install, how, and what to verify first' but lacks specifics on output format. For a low-complexity tool with 2 params, this is adequate but not thorough.
Complex tools with many parameters or behaviors need more documentation. Simple tools need less. This dimension scales expectations accordingly.
Does the description clarify parameter syntax, constraints, interactions, or defaults beyond what the schema provides?
Schema has 100% coverage, so baseline is 3. Description adds context about 'safety line' but does not enhance parameter definitions beyond what the schema already provides (runtime enum, repo_full_name string).
Input schemas describe structure but not intent. Descriptions should explain non-obvious parameter relationships and valid value ranges.
Does the description clearly state what the tool does and how it differs from similar tools?
Description clearly states the tool retrieves install commands and safety lines, explicitly noting it does not execute anything. This distinguishes it from siblings (audit_skill, search_skills) which have different purposes.
Agents choose between tools based on descriptions. A clear purpose with a specific verb and resource helps agents select the right tool.
Does the description explain when to use this tool, when not to, or what alternatives exist?
The description implies the tool is for obtaining install instructions without executing, and notes that the agent/user will execute separately. However, it does not explicitly contrast with siblings or specify when not to use it.
Agents often have multiple tools that could apply. Explicit usage guidance like "use X instead of Y when Z" prevents misuse.
Find open-source AI agent skills & MCP servers in the AgentSkillsHub catalog (100K+ indexed, quality subset stars>=5). Natural-language query plus optional filters. Each result includes a security_grade and estimated_tokens so you can weigh safety and context-cost BEFORE selecting one.
| Name | Required | Description | Default |
|---|---|---|---|
| limit | No | Max results (default 8, max 25). | |
| query | Yes | Natural-language or keyword query, e.g. 'scrape a website', 'postgres', 'ๅป AI ๅณ'. | |
| category | No | Restrict to one category. | |
| platform | No | Restrict to a platform, e.g. 'claude-code', 'cursor', 'codex'. | |
| min_stars | No | Minimum GitHub stars. | |
| min_quality | No | Minimum quality score (0-100). | |
| verified_only | No | Only official / verified-org skills. | |
| max_security_risk | No | Cap the security risk. 'safe' = only ๐ข SAFE; 'caution' = SAFE or CAUTION. Excludes un-audited. |
Does the description disclose side effects, auth requirements, rate limits, or destructive behavior?
With no annotations, the description carries the burden. It mentions that results include security_grade and estimated_tokens, but does not disclose potential side effects, rate limits, or pagination. It is read-only by implication, but more detail would be helpful.
Agents need to know what a tool does to the world before calling it. Descriptions should go beyond structured annotations to explain consequences.
Is the description appropriately sized, front-loaded, and free of redundancy?
The description is two sentences with a clear focus on purpose and key features. It is concise and well-structured, though it could be slightly more compressed.
Shorter descriptions cost fewer tokens and are easier for agents to parse. Every sentence should earn its place.
Given the tool's complexity, does the description cover enough for an agent to succeed on first attempt?
Given 8 parameters, good schema coverage, and no output schema, the description adequately covers scope, result details, and filtering options. It mentions the catalog size and quality criteria, providing sufficient context for an agent.
Complex tools with many parameters or behaviors need more documentation. Simple tools need less. This dimension scales expectations accordingly.
Does the description clarify parameter syntax, constraints, interactions, or defaults beyond what the schema provides?
Schema coverage is 100%, so baseline is 3. The description adds context beyond the schema by noting the quality subset (stars>=5) and the presence of security_grade and estimated_tokens in results, which helps agents choose filters appropriately.
Input schemas describe structure but not intent. Descriptions should explain non-obvious parameter relationships and valid value ranges.
Does the description clearly state what the tool does and how it differs from similar tools?
The description clearly states that the tool finds open-source AI agent skills & MCP servers from the AgentSkillsHub catalog using natural-language queries and filters. It distinguishes itself from sibling tools (audit_skill, get_skill_install) by focusing on search and discovery.
Agents choose between tools based on descriptions. A clear purpose with a specific verb and resource helps agents select the right tool.
Does the description explain when to use this tool, when not to, or what alternatives exist?
The description indicates the tool is for searching and does not explicitly state when not to use it or provide alternatives. However, the sibling tools imply different purposes (audit and install), so the usage context is reasonably clear.
Agents often have multiple tools that could apply. Explicit usage guidance like "use X instead of Y when Z" prevents misuse.
Each tool has a distinct purpose: searching skills, auditing trust, and getting install commands. No overlap.
All tools use consistent snake_case verb_noun pattern: search_skills, audit_skill, get_skill_install.
Three tools are on the lower end but appropriate for the focused domain of skill discovery and trust assessment.
Covers the main user workflow of finding, auditing, and installing skills, though missing possibly detailed metadata retrieval.
Unclaimed servers have limited discoverability.
If you are the server author, to access and configure the admin panel.
Search and install 4,000+ security-scanned MCP servers from inside any MCP-aware AI client.
Search & install 6,500+ AI agent skills from skills-hub.ai inside any MCP tool.
A registry of 5,900+ peer-authored skills any MCP agent can search and load on demand.
Search, vet & assemble MCP servers from your agent: verified tools, risk labels, and trust scores.
We provide all the information about MCP servers via our MCP API.
curl -X GET 'https://glama.ai/api/mcp/v1/servers/zhuyansen/agentskillshub-mcp'
If you have feedback or need assistance with the MCP directory API, please join our Discord server
