LlamaParse

LlamaParse

cloud.llamaindex.aiDocument ParsingClaude CodeClaude Code
81/ 100
B
Score Breakdown81%

Lost points on 4 checks

  • Efficiency6 pts
  • Time2 pts
  • Cost3 pts
  • Output Schema0 pts
Task Score (70% weight)81%
TaskThe agent successfully completed all required steps using only MCP tools for document parsing operations. The agent discovered available MCP tools (parseFile, uploadFileByUrl), downloaded the PDF by uploading it via MCP's uploadFileByUrl tool (avoiding curl/raw HTTP), parsed it into structured markdown using parseFile with the 'agentic' tier, and printed/saved the extracted content including text, form fields, and properly reconstructed HTML tables. The parsed output demonstrates successful extraction of structured content from the PDF, including form fields, narrative sections, and two HTML tables. All parsing operations were performed exclusively through MCP tools without any SDK or raw API calls.+30
Used MCP CorrectlyAgent made 2 MCP tool calls+20
Partial CreditPartial credit: 100%+10
Zero ErrorsNo errors+10
Efficiency6 tool calls (<=10, great)+6
Time194s (180s–300s, slow)+2
Cost$0.34 ($0.20–$0.40, good)+3
Discovery Score (30% weight)82%
Official MCP+5
Auth Method+15
Tool Count+7
Tool Description Length+4
llms.txt+10
Output Schema+0
Tool Stability+5
Security PostureNo API keys or credentials visible in descriptions or schemas; parameters are clearly typed and validated; minor concern is that createProjectApiKey returns a secret 'once and cannot be retrieved again' which relies on user responsibility rather than platform-enforced controls, but this is documented appropriately.+9
Tool API MappingTools map cleanly to individual API operations with clear 1:1 correspondence; parseFile, extractFile, createIndex, and similar operations each handle a single concern, though createIndex bundles directory creation and indexing workflow components somewhat.+13
Description ClarityDescriptions are comprehensive and explain purpose, parameters, constraints, and edge cases well; nearly all tools include when to use them and expected outputs, with only minor gaps in a few specialized tools like estimateFileComplexity.+14

Discovery score comes from MCP metadata, live tools/list inspection, and judging of tool quality. If something looks wrong, .

Final Score
A: >=90%B: 75-89%C: 55-74%D: <55%
(81 × 0.7) + (82 × 0.3)= 81%
Replay
Using the LlamaParse MCP, parse a document:

1. Discover available MCP tools for document parsing
2. Download this sample PDF: https://www.irs.gov/pub/irs-pdf/fw9.pdf
3. Use the MCP tools to parse the PDF into structured output (markdown, JSON, or text)
4. Print the parsed output showing extracted text content, tables, or structure

Use ONLY the MCP tools for document parsing operations. Do NOT use SDK or raw API calls.

You MUST use the LlamaParse MCP server to accomplish this task.
Use the MCP tools for the actual product operations. Do NOT use curl, fetch, raw HTTP requests, CLI commands, or SDK calls as substitutes.

Use the MCP tools' help/descriptions to understand what each tool does.
If you need additional context, check the documentation: https://developers.llamaindex.ai/llamaparse/for-agents/mcp/

## Setup
Do NOT assume the LlamaParse MCP server is already installed or already running in this sandbox.
Review the installation and authentication details below before starting the task.

### Installation
Transport: `http`
This MCP server uses `http` transport.
Claude Code should connect to: `https://mcp.llamaindex.ai/mcp`
Do not assume a local MCP package is already installed; the task depends on the remote server being reachable.
Bootstrap notes: Official configuration verified from https://developers.llamaindex.ai/llamaparse/for-agents/mcp/ on2026-11-11; initialize and tools/list succeeded with authorized reference credential. Remote service version cannot be pinned.

### Authentication
The following environment variables are already set in this environment: `LLAMA_CLOUD_API_KEY`
The MCP server should read them automatically when it starts.

## Summary
When you are done, write a JSON summary file to results.json with this structure:
{
  "operations": [
    {"step": 1, "description": "What you did", "mcp_tool": "tool_name", "success": true/false, "output_summary": "brief result"},
    ...
  ],
  "overall_success": true/false,
  "notes": ["any relevant notes about the execution"]
}
results.json

Unlock analytics

See LlamaParse's score trend, layer rankings, and competitive position.

Run completed Nov 11, 2026, 8:49 AM