agent-tools — independently scanned and version-tracked by SaferSkills.
SaferSkills independently audited agent-tools (Agent Skill) and scored it 100/100 (green). The audit ran 55 deterministic rules across Security, Supply Chain, Maintenance, Transparency, and Community; it found 0 high-severity and 0 lower-severity findings. The full rule-by-rule trace and per-finding evidence are below. Free, methodology-open.
Findings & checks · 0 flagged
Every scanned point with the score it earned and what moved between them.
First recorded scan — no prior version to compare against.
The primary manifest — the file an agent reads to learn what this artifact does.
Discovery layer for the agent economy. Lets an agent find an agent-callable resource by intent, read its full call template, and use it.
Backed by agent-tools.cloud, an open directory that indexes three kinds of resources:
| Resource | What it solves | Typical object |
|---|---|---|
| x402 service | pay-per-call HTTP API (USDC on Base) | /.well-known/x402, 402 endpoint |
| MCP server | external tools & context via MCP | Streamable-HTTP / stdio MCP servers |
| A2A agent | delegate a task to another agent | /.well-known/agent-card.json, A2A JSON-RPC |
A2A tells agents who can do the job. x402 lets them pay for it. MCP gives them the tools to do it.
Sources include awesome-x402, x402scan, x402.org/ecosystem, the MCP registry / PulseMCP, and public A2A agent-cards, plus a handful of self-hosted services on the same host.
Use this skill when the user asks for any of:
specific provider — search first, then decide.
Do not use this skill for:
The MCP server (installed via the installer specs above) exposes these tools:
| Tool | Purpose |
|---|---|
search(intent, top_k?, max_price_usd?, category?) | Find x402 paid services |
get(slug) | Full x402 record (URL, schema, price, x402 bazaar info) |
list_categories() | Browse x402 categories |
search_mcp_servers(intent, top_k?, chain?) | Find MCP servers |
get_mcp_server(slug) | Full MCP server record (endpoint, transport, capabilities) |
search_agents(intent, top_k?, x402_only?) | Find A2A agents |
get_agent(slug) | Full A2A agent card (endpoint, skills, auth, x402 info) |
search_all(intent, protocol?, top_k?) | Unified search across all three; each result tagged with protocol |
stats() | Live directory size + health across x402 / MCP / A2A |
If you don't know which resource type fits, start with search_all. Otherwise use the protocol-specific search, pick the highest-ranked result that fits the budget, then get*(slug) for the full call template.
For x402 services hosted on agent-tools.cloud, get() includes the full x402 v2 extensions.bazaar block — request body example, JSON Schema, output example — so the agent can construct a request without trial-and-error. MCP servers return their endpoint URL + transport; A2A agents return their agent-card (skills, endpoint, auth). Third-party entries are passed through as scraped; when in doubt, call the endpoint with an empty body and read the 402 challenge (x402) or fetch the agent-card / MCP tools/list for guidance.
payment-requiredheader (base64-encoded x402 v2 challenge).
X-Payment header.Use any x402-compatible payment lib (x402-axios, x402-fetch, the Python x402.payment helpers, …) to handle steps 1–3. MCP servers and A2A agents are connected/called via their own protocol, not x402 (unless flagged x402-capable).
Set in ~/.openclaw/openclaw.json (optional):
{
"skills": {
"entries": {
"agent-tools": {
"enabled": true,
"env": {
"AGENT_TOOLS_API_BASE": "https://agent-tools.cloud"
}
}
}
}
}AGENT_TOOLS_API_BASE lets you point at a self-hosted directory if you ever deploy your own.
The directory itself is free (all search*, get*, list_categories, stats are unauthenticated, no rate limit announced). Only the underlying paid x402 services charge — typically $0.001–$0.50 per call. MCP servers and A2A agents have their own pricing (many are free).
The agent / user is in full control of which 402 challenge to actually settle.
price_usd to the user before paying.max_price_usd in search() if the user mentioned a budget.~30 seconds. Free. No account. Every finding cites a rule and a line of evidence.