ComPDFSAFE
A lightweight MCP (Model Context Protocol) server for integrating ComPDF AI with Claude Desktop, enabling AI-powered intelligent document processing and data extraction from PDFs via natural language.
Overview
From the repository's own README, as read at the audited commit. Badges and raw HTML are left out.
English | 繁體中文 | 日本語 | 简体中文
ComPDF MCP is built for MCP clients and AI agent platforms. It provides PDF/image parsing and data extraction, PDF conversion, PDF editing, and image conversion capabilities. It helps users process raw documents first, then pass lighter, cleaner, and more structured results into AI, reducing token usage, lowering AI costs, and improving overall processing efficiency.
If you find ComPDF MCP useful, please consider giving us a ⭐ Star on GitHub. It helps us grow and improve. Got questions or ideas? Join the conversation in our Discussions.
Why Choose ComPDF MCP • Capabilities • Technical Architecture • Quick Start • Deployment and Operations • License and Free Access • Use Cases and Example Prompts • Support
Why Choose ComPDF MCP
The core reason users choose ComPDF MCP is not simply to let AI work with PDFs. It is to process complex, lengthy, and noisy source documents into inputs that are easier for models to understand before handing them off to AI for analysis, summarization, Q&A, or automation.
Raw PDFs, scanned files, and image
41594ecfd605OBSERVED · 2026-10-07Connect
Built from this server's own package name, version and transport as found in its source — not copied from anyone's documentation, so it cannot drift against a page we do not control. Replace the environment placeholders with a token scoped to the least it needs.
claude mcp add compdf-mcp --env MCP_OAUTH_ISSUER_URL=${MCP_OAUTH_ISSUER_URL} --env MCP_STATIC_TOKENS_JSON=${MCP_STATIC_TOKENS_JSON} -- uvx compdf-mcp{
"mcpServers": {
"compdf-mcp": {
"command": "uvx",
"args": [
"compdf-mcp"
],
"env": {
"MCP_OAUTH_ISSUER_URL": "${MCP_OAUTH_ISSUER_URL}",
"MCP_STATIC_TOKENS_JSON": "${MCP_STATIC_TOKENS_JSON}"
}
}
}
}Exposed tools (6)
2 read · 4 write · 0 destructive.
| Tool | Risk | Description |
|---|---|---|
create_presigned_upload | write | try: |
get_task_status | read | try: |
list_operations | read | return [_operation_summary(operation) for operation in operations] |
start_async_operation | write | try: |
start_presigned_operation | write | if task_id not in presigned_jobs: |
upload_presigned_file | write | job = presigned_jobs.get(task_id) |
Trust audit
SAFEgrade B · trust 89/100 Nothing in the source contradicts what it says it does. Grade A is reserved for packages that have also passed the behavioural sandbox.
| Layer | What it checks | Result |
|---|---|---|
| L0 | Provenance & inventory | PASS |
| L1 | Static analysis of the code | WARN |
| L2 | Instruction surface (what it tells the agent) | PASS |
| L3 | Class-specific surface | PASS |
| L4 | Behavioural (sandbox) | SKIPPED |
What the source does
- Filesystem
- declared (1 observation(s))
- Network
- declared (5 observation(s))
- Shell
- none-observed
- Dependencies
- pinned
- Secrets in source
- none-found
Findings (12)
CMD python -c "from urllib.request import urlopen; urlopen('http://127.0.0.1:8000/healthz', timeout=3)""description": "Encryption algorithm. Allowed values: `rc4`, `aes128`, `aes256`. Default: `rc4`.",
"default": "rc4",
"rc4",
"example": "rc4"
http://127.0.0.1:8000/mcp
The examples below use the unified MCP URL `http://127.0.0.1:8000/mcp`. For a remote deployment, replace it with the public HTTPS address, for example `https://your-domain.example/mcp`.
url = "http://127.0.0.1:8000/mcp"
"url": "http://127.0.0.1:8000/mcp",
content = base64.b64decode(encoded, validate=True)
In production, expose the aggregate endpoint as `https://your-domain.example/mcp` behind an HTTPS reverse proxy, and configure `MCP_ENV=production`, `MCP_PUBLIC_URL`, `MCP_OAUTH_ISSUER_URL`, and `MCP_
assets/ComPDF-Comprehensive PDF Solutions.png
Gates applied: no_behavioural_pass.
41594ecfd605full audit observations/trust-audit/mcp-server/compdfkit__compdf.json · Report an issue / request a re-scanAudit history
Every audit this server has had. A grade with a past is a grade somebody is still checking.
| Date | Source | Verdict | Grade | Score | Change |
|---|---|---|---|---|---|
| 2026-10-07 | 41594ecfd605 | SAFE | B | 89 | first audit |
Questions
What is the ComPDF MCP server?
A lightweight MCP (Model Context Protocol) server for integrating ComPDF AI with Claude Desktop, enabling AI-powered intelligent document processing and data extraction from PDFs via natural language.
What tools does ComPDF expose?
6 in total: 2 read-only, 4 that write, and 0 that can delete or overwrite. Every one is listed on this page with its risk.
Is ComPDF safe to connect to an agent?
The audit found nothing in the source that contradicts what it says it does, and graded it B (89/100). Grade A is held back for packages that have also passed a sandboxed behavioural run, which is why a clean server reads B.
What credentials does ComPDF need?
It reads MCP_OAUTH_ISSUER_URL and MCP_STATIC_TOKENS_JSON from the environment. Give it a token scoped to the least it needs — an agent that can be talked into calling a tool can be talked into calling it with your credentials.
How does ComPDF run?
It speaks streamable-http, so it runs as a service you connect to over the network. It is published on PyPI as compdf-mcp.
How current is this page?
The grade is for one exact copy of the source (41594ecfd605), read on 2026-10-07. The repository is watched and re-audited when it changes.