Ucm

PassAudited by ClawScan on May 10, 2026.

Overview

UCM is a coherent API-marketplace skill, but it can make paid external API calls and handle a UCM API key, so users should monitor credits and data sent to providers.

Before installing, decide whether you want an agent to access UCM's broad external API marketplace. Monitor credit usage, ask the agent to confirm paid or sensitive calls, avoid sending secrets or confidential files/content, and keep the UCM_API_KEY private.

Findings (4)

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.

What this means

The agent may make external API calls and spend UCM credits when using this skill.

Why it was flagged

The skill intentionally gives the agent curl-based access to UCM marketplace calls that may consume credits. This is disclosed and central to the purpose, but it is broad tool authority.

Skill content
allowed-tools: Bash(curl:*) Grep ... instantly discover and use API services by spending credits
Recommendation

Use the skill only if you are comfortable with agent-initiated marketplace calls; monitor usage and ask for confirmation before paid or sensitive operations if desired.

What this means

Anyone who can read the config file or terminal logs may be able to use the UCM API key and spend the associated credits.

Why it was flagged

The helper script stores the UCM API key locally and prints it to the terminal. This is expected for registration, but it is sensitive credential material.

Skill content
Credentials are saved to ~/.config/ucm/credentials.json ... echo "API Key: $API_KEY"
Recommendation

Protect the credentials file, avoid sharing logs that contain the API key, and rotate the key if it is exposed.

What this means

Code submitted through the skill may run in UCM's remote sandbox and could expose any secrets included in that code or input.

Why it was flagged

The marketplace includes a remote code-execution service. The catalog describes it as sandboxed and fresh per execution, making it purpose-aligned but still worth user awareness.

Skill content
`ucm/code-sandbox` | $0.03 | 1 | Execute Python/JS/Bash/R/Java in sandbox
Recommendation

Do not send private keys, tokens, or confidential data to the code sandbox unless you intend to share it with that service.

What this means

Information sent through UCM may be processed by UCM and downstream providers rather than staying local.

Why it was flagged

UCM acts as a gateway to external providers and may receive user-supplied content such as queries, URLs, audio, prompts, or email bodies.

Skill content
Real-time web search powered by Tavily ... `audio_base64` OR `audio_url` (one required): Audio input
Recommendation

Avoid sending sensitive personal, business, or credential data through marketplace services unless you have reviewed and accept the provider's privacy terms.