DocuClaw

Sovereign document intelligence & archival system. Extracts structured data from invoices, receipts, and contracts 100% locally using AI.

Security Scan

The skill's claims (100% local processing, CLI command, and support for cloud/back-end models) don't match the packaged content — there is no binary, install, or credential guidance, and the runtime instructions are vague, giving the agent broad discretion.

Purpose & Capability

The README-style description promises a CLI workflow (e.g., `docuclaw process`), local-only processing, and support for various model backends (Ollama, OpenAI Vision). But the package contains no binary, no install instructions, and no environment/credentials. A legitimate document-extraction tool would need either an installable binary/library or clear instructions and (if using cloud models) API keys — none are present.

Instruction Scope

SKILL.md tells the agent to run a CLI command and to 'sync' extracted data to calendars/accounting tools, but gives no concrete endpoints, auth guidance, or safe limits. The instructions are high-level and open-ended, which grants the agent broad discretion to choose tools, network endpoints, or secrets to accomplish integration — more scope than the skill declares.

Install Mechanism

There is no install spec (lowest-risk pattern), but that's inconsistent with the instruction to run a `docuclaw` CLI. Either an install step or bundled code is missing. The lack of installation artifacts makes the skill non-functional as-is and ambiguous about what would be installed or run.

Credentials

The skill claims support for OpenAI Vision (a cloud service) and other backends but declares no required environment variables or primary credentials. If cloud models are used, API keys would be required; if truly '100% local' is intended, there should be explicit instructions for local model setup (e.g., Ollama). The absence of either is disproportionate to the stated capabilities.

Persistence & Privilege

The skill does not request always:true or other elevated persistence. It is user-invocable and can be called by the agent as normal, which is expected for skills. No evidence it attempts to modify other skills or global agent configuration.

What to consider before installing

This skill is incomplete and inconsistent. Before installing or enabling it, ask the maintainer for: (1) the actual executable or install instructions (how to get `docuclaw`), (2) clear, concrete instructions for using local models (how to set up Ollama) and/or what cloud APIs it uses and the exact environment variables it needs, (3) where and how data is stored on disk (paths, retention), and (4) what endpoints are used for 'sync' to calendars/accounting. Do not provide API keys or other credentials until those details are clear. If you need true local-only processing, prefer a skill with bundled code or verifiable install steps and explicit local model instructions; otherwise treat this skill as non-functional or potentially misleading.

DocuClaw Skill

DocuClaw provides a sovereign data infrastructure for processing and archiving documents. It uses multimodal LLMs to extract structured information from scans, photos, and emails, storing everything in human-readable, version-controllable Markdown files.

Use Cases

• Expense Management: Extract totals, taxes, and dates from receipts for tax filing.
• Contract Analysis: Monitor expiration dates and renewal clauses in legal documents.
• Sovereign Archival: Maintain a local-first, GDPR/GoBD compliant archive of all physical and digital mail.
• Unified Querying: Ask questions about your document history without cloud exposure.

Key Features

• 100% Local: Zero cloud dependency. Your private data never leaves your hardware.
• Plug-and-Play Parsers: Extensible architecture for country-specific document formats.
• AI-Powered: Supports Ollama, OpenAI Vision, or any multimodal model for intelligent extraction.
• Markdown Schema: Normalizes all documents into a universal schema with YAML metadata.

Workflow Example

1. Input: A PDF invoice or a photo of a receipt.
2. Process: Run docuclaw process to trigger AI extraction.
3. Archive: Document is saved to your local vault as YYYY/MM/filename.md.
4. Action: The extracted data is synced to your calendar or accounting tool.

Integration

DocuClaw is designed to work seamlessly with the OpenClaw ecosystem, allowing AI agents to perform RAG (Retrieval-Augmented Generation) over your local document archive.