Skillboss

ReviewAudited by ClawScan on May 10, 2026.

Overview

This appears to be a real AI API gateway, but it exposes raw API keys and includes email/SMS batch-sending capabilities without visible safeguards.

Install only if you trust SkillBoss with the prompts, files, and API key you provide. Avoid using the email/SMS models unless you require them, require explicit approval before any send or batch send, and keep setup logs containing API keys 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

A SkillBoss API key could be captured in shell history, logs, or agent transcripts, allowing someone else to use account credits or access the service as the user.

Why it was flagged

The setup flow saves a returned API key locally and prints the full key to stdout, rather than only masking it.

Skill content
config.apiKey = data.api_key; saveConfig(config); ... process.stdout.write(data.api_key);
Recommendation

Treat setup output as sensitive, avoid sharing logs, rotate the key if exposed, and prefer a secure environment variable or a helper that only prints keys with explicit user intent.

What this means

If invoked incorrectly or by an over-eager agent, the skill could send unwanted emails or SMS messages and potentially incur costs or reputational harm.

Why it was flagged

The skill documents external messaging actions, including batch email and SMS sends, but the artifacts do not show recipient review, allowlists, rate limits, or confirmation requirements.

Skill content
`email/send` | Send single email ... `email/batch` | Send batch emails ... `prelude/notify-batch` | Batch SMS notifications
Recommendation

Only enable messaging models when needed, require explicit user confirmation with recipient/message review before sending, and add allowlists or batch-size limits.

What this means

Prompts, context, and selected files may leave the local machine and be processed by SkillBoss and its downstream model providers.

Why it was flagged

For speech-to-text, a user-selected local file is encoded and sent to the remote SkillBoss API.

Skill content
inputs.audio_data = readFileSync(filePath).toString("base64"); ... body: JSON.stringify(body)
Recommendation

Do not submit confidential files or prompts unless the user accepts the provider data flow; choose specific models instead of smart routing when provider choice matters.

What this means

Users have less independent provenance information for verifying the publisher and service relationship.

Why it was flagged

The registry metadata lacks a source repository and homepage, even though the skill contacts external service domains.

Skill content
Source: unknown; Homepage: none
Recommendation

Review the included scripts and service domains before use, and prefer installing from publishers with clear source and support links.