← Back to Skills Marketplace
1782
Downloads
5
Stars
6
Active Installs
5
Versions
Install in OpenClaw
/install mistral-ocr
Description
Extract text, tables, and images from PDFs or images using Mistral OCR API and output in Markdown, JSON, or HTML formats.
Usage Guidance
This skill will upload any file you give it to Mistral's servers for OCR. Only install/use it for non-sensitive documents unless you accept that remote processing will occur and you've reviewed Mistral's privacy policy. Installing requires the mistralai Python package (pip install -r requirements.txt) and setting MISTRAL_API_KEY in your environment; limit the key's scope where possible and rotate/revoke it if you stop using the skill. Minor note: the package metadata in _meta.json shows a different ownerId than the registry header—not necessarily malicious but worth verifying the repository/source if provenance matters. If you need offline processing for confidential documents, use a local OCR tool instead.
Capability Analysis
Type: OpenClaw Skill
Name: mistral-ocr
Version: 1.0.4
The skill transparently uploads user-provided files to Mistral's cloud servers for OCR processing, as explicitly stated in the 'Privacy Warning' sections of both SKILL.md and README.md. The `scripts/mistral_ocr.py` code uses the official `mistralai` client library to perform this action, retrieving the `MISTRAL_API_KEY` from environment variables. There is no evidence of unauthorized data exfiltration, malicious execution, persistence mechanisms, prompt injection attempts against the agent, or obfuscation. All actions align with the skill's stated purpose and are clearly disclosed to the user.
Capability Assessment
Purpose & Capability
Name/description (Mistral OCR) align with required artifacts: the SKILL.md, README, requirements.txt and script all declare and use MISTRAL_API_KEY and the mistralai client to upload files for OCR. There are no unrelated credentials or binaries requested.
Instruction Scope
SKILL.md explicitly instructs uploading files to Mistral and warns about privacy. The included script only reads the user-specified input file, uses MISTRAL_API_KEY, uploads the file to Mistral, calls the OCR model, and writes output and extracted images locally. The instructions do not request additional system files, secrets, or unexpected external endpoints.
Install Mechanism
There is no automated install spec (instruction-only), and dependencies are limited to a single Python package (requirements.txt: mistralai). This is proportionate for a Python CLI tool that talks to Mistral. Installing from PyPI (mistralai) is the expected setup; no opaque downloads or extract-from-URL steps are present.
Credentials
The skill requires a single environment variable (MISTRAL_API_KEY) which is necessary and sufficient for the declared cloud OCR functionality. The code does not read additional environment variables or other credentials.
Persistence & Privilege
The skill is not always-enabled, does not request persistent platform privileges, and does not modify other skills or system-wide agent settings. It runs as a normal user-mode CLI script.
How to Use
- Make sure OpenClaw is installed (local or Docker)
- Run the install command in chat:
/install mistral-ocr - After installation, invoke the skill by name or use
/mistral-ocr - Provide required inputs per the skill's parameter spec and get structured output
Version History
v1.0.4
Version 1.0.4 of mistral-ocr
- No file changes were detected in this release.
- No updates to features, documentation, or functionality.
v1.0.3
- Added comprehensive privacy warning about uploading files to Mistral cloud servers.
- Expanded documentation: added installation steps, usage examples, and detailed API key setup in SKILL.md.
- Enhanced feature description to highlight support for multi-column layouts and provide more usage details.
- Updated metadata to include homepage, author, and required environment variables.
- Replaced skill.md with new SKILL.md; added README.md and _meta.json.
v1.0.2
- Improved documentation: expanded SKILL.md with detailed usage instructions, CLI examples, supported formats, and output information.
- Clarified API key requirement and environment variable setup for easier configuration.
- Added table of CLI arguments for reference and usability.
- Included examples for various output formats (Markdown, JSON, HTML) and image extraction details.
v1.0.1
Remove test output files, keep only clean skill files
v1.0.0
Initial release of mistral-ocr.
- Converts PDF and common image formats to Markdown, JSON, or HTML using the Mistral OCR API.
- Supports image extraction, table recognition, and header/footer handling.
- Offers both file upload and CLI usage.
- Outputs Markdown, structured JSON data, and extracted images.
- Requires the MISTRAL_API_KEY environment variable for authentication.
Metadata
Frequently Asked Questions
What is Mistral OCR?
Extract text, tables, and images from PDFs or images using Mistral OCR API and output in Markdown, JSON, or HTML formats. It is an AI Agent Skill for Claude Code / OpenClaw, with 1782 downloads so far.
How do I install Mistral OCR?
Run "/install mistral-ocr" in the OpenClaw or Claude Code chat to install it in one step — no extra setup required.
Is Mistral OCR free?
Yes, Mistral OCR is completely free (open-source). You can download, install and use it at no cost.
Which platforms does Mistral OCR support?
Mistral OCR is cross-platform and runs anywhere OpenClaw / Claude Code is available (cross-platform).
Who created Mistral OCR?
It is built and maintained by LeyuDame2 (@yzdame); the current version is v1.0.4.
More Skills