Upstage Information Extraction

PassAudited by ClawScan on May 6, 2026.

Overview

This skill appears to do what it says, but it needs an Upstage API key and may send or store sensitive document data during extraction.

Before installing, confirm you are comfortable sending the relevant documents or document URLs to Upstage, store the UPSTAGE_API_KEY securely, and clean up any temporary extracted JSON files that contain sensitive data.

Findings (3)

Artifact-based informational review of SKILL.md, metadata, install specs, static scan signals, and capability signals. ClawScan does not execute the skill or run runtime probes.

Note

ASI03: Identity and Privilege Abuse

What this means

A user must provide an Upstage API key, which can authorize API usage and may incur account usage or billing depending on the Upstage account.

Why it was flagged

The skill requires a sensitive Upstage API credential. This is expected for the service integration, but users should notice that a credential is needed.

Skill content

API Key: Always use `os.environ["UPSTAGE_API_KEY"]`.

Recommendation

Use a dedicated Upstage key with the minimum needed scope if available, keep it in environment variables, and rotate it if it is exposed.

Note

ASI07: Insecure Inter-Agent Communication

What this means

Documents or document URLs provided to the skill may be processed by Upstage, so sensitive content may leave the user's local environment.

Why it was flagged

The extraction workflow sends document references to Upstage's external API. This is central to the skill, but invoices, receipts, and trade documents can contain sensitive business or personal data.

Skill content

base_url="https://api.upstage.ai/v1/information-extraction" ... "image_url": {"url": "https://example.com/invoice.pdf"}

Recommendation

Only use the skill for documents you are allowed to send to Upstage, and review Upstage's data retention and privacy terms for sensitive files.

Note

ASI06: Memory and Context Poisoning

What this means

Sensitive extracted data may remain on disk after the task completes, depending on temp-directory cleanup behavior.

Why it was flagged

The skill stores extracted structured data in a local temp file by default. This is purpose-aligned, but extracted fields may include sensitive financial or identity information.

Skill content

Default: write extracted JSON to `<system-temp>/<input-stem>.extracted.json` (e.g., `/tmp/invoice.extracted.json`).

Recommendation

Choose an explicit output path when needed, protect the file, and delete temporary extraction outputs when they are no longer needed.