Skip to main content
Glama
jnilrac
by jnilrac
README.md
# Secant Agent Research MCP

[![CI](https://github.com/jnilrac/secant-agent-research-mcp/actions/workflows/ci.yml/badge.svg)](https://github.com/jnilrac/secant-agent-research-mcp/actions/workflows/ci.yml)
[![secant-agent-research-mcp MCP server](https://glama.ai/mcp/servers/jnilrac/secant-agent-research-mcp/badges/score.svg)](https://glama.ai/mcp/servers/jnilrac/secant-agent-research-mcp)
[![secant-agent-research-mcp MCP server](https://glama.ai/mcp/servers/jnilrac/secant-agent-research-mcp/badges/card.svg)](https://glama.ai/mcp/servers/jnilrac/secant-agent-research-mcp)
[![smithery badge](https://smithery.ai/badge/hello-j2mo/secant-agent-research-pack)](https://smithery.ai/servers/hello-j2mo/secant-agent-research-pack)

MCP wrapper for the [Secant Agent Research Pack](https://agentic.secantoutreach.com/agent-research), a paid x402 research API for autonomous agents.

The server exposes tools for:

- `search`: ranked web search.
- `research_pack`: search, selected page extraction, normalized JSON, and citations.
- `extract_page`: readable page text and metadata extraction.
- `monitor_diff`: page change checks.

## Tool Behavior

All tools use the same retry-safe x402 pattern:

1. Call the tool without `payment` to receive the upstream payment requirement.
2. Pay the quoted Base USDC requirement with an x402-compatible client.
3. Retry with the payment proof and, when available, the same `payment_identifier`.

The wrapper returns `payment_required: true` and the upstream 402 body for unpaid calls. It does not execute paid work until the live API accepts the payment proof.

The native paid API uses HTTP `402` and Base USDC over x402. If you call a tool without a payment proof, the tool returns the upstream payment requirement so an agent can pay and retry. If you include a valid payment proof, the tool forwards it to the live API.

## Run

```bash
docker build -t secant-agent-research-mcp .
docker run --rm -i secant-agent-research-mcp
```

## Environment

| Variable | Default | Description |
| --- | --- | --- |
| `SECANT_AGENT_RESEARCH_API_BASE_URL` | `https://agentic.secantoutreach.com` | Base URL for the live Secant Agent Research API. |

## Canonical Discovery

- Buyer page: <https://agentic.secantoutreach.com/agent-research>
- Smithery listing: <https://smithery.ai/servers/hello-j2mo/secant-agent-research-pack>
- x402 manifest: <https://agentic.secantoutreach.com/.well-known/x402.json>
- OpenAPI: <https://agentic.secantoutreach.com/openapi.yaml>
- Agent card: <https://agentic.secantoutreach.com/agent-card.json>

TDQS

A4.1/5.0

Scored across 4 tools

Disambiguation4/5

The tools are distinct in purpose: search for web searching, extract_page for content from known URLs, monitor_diff for change detection, and research_pack as a higher-level combined workflow. While research_pack encompasses search and extraction, it is presented as a different service level, so confusion is minimal.

Naming Consistency4/5

Three tools follow a verb_noun pattern (extract_page, monitor_diff, research_pack), with 'research_pack' being noun_noun but still clear. 'search' is a single verb, slightly deviating but acceptable. Overall pattern is recognizable and consistent.

Tool Count5/5

Four tools is an appropriate number for a research-focused server, covering key operations (search, extraction, monitoring, and a bundled paid workflow) without being overbearing or insufficient.

Completeness4/5

Core research workflows (search, extract, monitor, full research pack) are covered. Minor gaps exist, such as lacking a tool to list or retrieve previous research packs, but the set is functional for autonomous agents.

Maintenance

ActivityInactive
ResponsivenessNo issues