skills/kompressrcdn/clients/kompressr/Gen Agent Trust Hub

kompressr

Pass

Audited by Gen Agent Trust Hub on Oct 19, 2026

Risk Level: SAFEEXTERNAL_DOWNLOADSCOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTIONDYNAMIC_EXECUTION
Full Analysis
  • [EXTERNAL_DOWNLOADS]: The skill provides instructions for installing official Kompressr SDKs from standard package registries (npm, PyPI, Maven, NuGet, Packagist, and RubyGems). These packages are vendor-owned resources used for the skill's primary purpose.
  • Evidence: Installation commands like npm install @kompressr/sdk (references/sdk-js.md), python -m pip install kompressr-sdk (references/sdk-python.md), and go get github.com/kompressrcdn/clients/sdk-go@v0.1.0 (references/sdk-go.md).
  • [COMMAND_EXECUTION]: The documentation includes curl command examples to demonstrate how to interact with the Kompressr HTTP API. These examples are designed for developer testing and rely on secure environment variable configuration for authentication.
  • Evidence: curl commands in references/http-api.md for uploading files, checking status, and signing private assets.
  • [DYNAMIC_EXECUTION]: The MCP integration guide describes configuring a host environment to execute the @kompressr/mcp-server package using npx. This facilitates local execution of vendor-provided tool logic for agent hosts.
  • Evidence: MCP server configuration block in references/mcp.md using "command": "npx", "args": ["-y", "@kompressr/mcp-server@2.0.2"].
  • [INDIRECT_PROMPT_INJECTION]: The skill's core functionality involves processing external files and metadata provided by the user to be uploaded to the Kompressr service. This introduces an attack surface where untrusted data could potentially contain instructions intended for an agent.
  • Ingestion points: External files and JSON metadata are ingested through upload_file, upload_bytes, and multipart API fields (SKILL.md, references/http-api.md).
  • Boundary markers: The skill includes instructions to validate file visibility, metadata types, and user authentication before dispatching data (references/custom-integration.md).
  • Capability inventory: The skill performs network operations to vendor origins (uploads.kompressr.com, kompressr.com) and reads local files for upload.
  • Sanitization: Developers are instructed to enforce scalar metadata limits, validate filenames, and use structured error handling to prevent leaking sensitive information (references/custom-integration.md).
Audit Metadata
Risk Level
SAFE
Analyzed
Oct 19, 2026, 06:16 PM