# Claude Skill Indexedai

> A Claude skill from guidodr27/claude-skill-indexedai.

- **Type:** Skill
- **Install:** `agentstack add skill-guidodr27-claude-skill-indexedai-claude-skill-indexedai`
- **Verified:** Yes — security-reviewed for prompt injection and unsafe behavior
- **Seller:** [guidodr27](https://agentstack.voostack.com/s/guidodr27)
- **Installs:** 0
- **Category:** [Agent Skills](https://agentstack.voostack.com/c/agent-skills)
- **Latest version:** 0.1.0
- **License:** MIT
- **Upstream author:** [guidodr27](https://github.com/guidodr27)
- **Source:** https://github.com/guidodr27/claude-skill-indexedai

## Install

```sh
agentstack add skill-guidodr27-claude-skill-indexedai-claude-skill-indexedai
```

Requires the [AgentStack CLI](https://agentstack.voostack.com/docs/cli). Works with Claude Code, Cursor, and any MCP-compatible agent.

## About

# Skill: /indexedai

Query any website via its MCP endpoint — structured data, zero context overhead. MCP loads only when invoked, never at startup.

## Trigger
- `/indexedai  [question]` — query a specific site
- `/indexedai  [question]` — domain-only shorthand
- `/indexedai ` — when a URL is implied by context

## Flow obbligatorio

Before fetching anything, tell the user:
> "Checking for MCP server at ``…"

### Step 1 — Discover MCP via `.well-known`

```
GET https:///.well-known/mcp.json
```

**Success** — response contains:
```json
{"mcpServers":{"name":{"url":"https://..."}}}
```
→ extract MCP endpoint URL, proceed to Step 2.

**Failure** (404, timeout >3s, malformed JSON, network error) → skip to **Fallback**.

### Step 2 — Initialize (mandatory handshake)

```
POST 
Content-Type: application/json

{"jsonrpc":"2.0","id":1,"method":"initialize","params":{"protocolVersion":"2024-11-05","clientInfo":{"name":"indexedai-skill","version":"1.0"}}}
```

If response missing `result.capabilities` or status ≥ 400 → skip to **Fallback**.

### Step 3 — List available tools

```
POST 
Content-Type: application/json

{"jsonrpc":"2.0","id":2,"method":"tools/list","params":{}}
```

Read `result.tools[]`. Pick the tool whose name/description best matches the user's question.
If no tools returned → skip to **Fallback**.

### Step 4 — Call the best tool

```
POST 
Content-Type: application/json

{"jsonrpc":"2.0","id":3,"method":"tools/call","params":{"name":"","arguments":{...}}}
```

Map user's question to tool arguments. Return `result.content`.

### Fallback

Tell the user:
> "No MCP server found at ``, using WebFetch."

Then use `WebFetch` on the most relevant URL for the question.

## Multiple domains

If query contains multiple domains, run Step 1 for **all** domains in parallel before calling any tool.
Then call each MCP concurrently.

## Output format

Always prefix answers with source:
- `[MCP: domain.com]` — data from site's own MCP server
- `[Web: domain.com]` — fallback WebFetch

Example:
```
[MCP: docs.stripe.com]
Webhooks retry up to 3 days with exponential backoff...
```

## Implementation notes

- HTTP stateless — no persistent connection to maintain
- No SDK or library needed — plain JSON-RPC over HTTP POST
- If `initialize` succeeds but `tools/list` fails → Fallback
- Never invent tool names — only use names returned by `tools/list`
- Respect `result.content[].type`: `text` = display as-is, `resource` = fetch if needed

## Source & license

This open-source skill is cataloged on AgentStack and links to its original source — we do not rehost the code.

- **Author:** [guidodr27](https://github.com/guidodr27)
- **Source:** [guidodr27/claude-skill-indexedai](https://github.com/guidodr27/claude-skill-indexedai)
- **License:** MIT

Install and usage instructions live in the source repository linked above.

## Pricing

- **Free** — Free

## Security capabilities

Automated source analysis of v0.1.0 — what this tool can access:

- **Network access:** no
- **Filesystem access:** no
- **Shell / process execution:** no
- **Environment & secrets:** no
- **Dynamic code execution:** no

*"Yes" means the capability is present in the source — more access means more to trust, not that it is unsafe.*


## Versions

- **0.1.0** — security scan: passed — Imported from the upstream source.

## Links

- Listing page: https://agentstack.voostack.com/l/skill-guidodr27-claude-skill-indexedai-claude-skill-indexedai
- Seller: https://agentstack.voostack.com/s/guidodr27
- Browse the marketplace: https://agentstack.voostack.com/browse

---
Listed on AgentStack — the marketplace for AI agent skills and MCP servers. Every listing is security-reviewed. Creators keep 70%.