SkillHub Manager

Manage and publish agent skills on SkillHub and ClawHub. Best for developers and operators who need a repeatable workflow to search skills, inspect packages,...

Security Scan

The skill's instructions match its stated purpose (using the clawhub CLI to inspect and publish skills) but the package metadata omits important operational details (not declaring npx as a required binary and not listing environment variables the instructions reference), which is inconsistent and could lead to accidental credential exposure or unexpected local file access.

Purpose & Capability

The name/description (manage and publish skills via SkillHub/ClawHub) aligns with the SKILL.md which instructs using the 'npx clawhub' toolchain to search, inspect, and publish. That capability legitimately requires a tool like 'npx' / Node/npm and the ability to read local skill folders and, when publishing, to use an API token. However, the skill metadata declares no required binaries or env vars while the runtime instructions clearly depend on 'npx' and token-related environment variables — an omission that makes the package metadata incomplete.

Instruction Scope

SKILL.md explicitly instructs the agent to: ask for and confirm a registry address, run 'npx clawhub' commands, validate existing tokens with 'npx clawhub whoami', read and publish a local folder (e.g., ./my-skill), and accept tokens from the user or environment. These actions (reading local directories, invoking npx, handling API tokens) go beyond passive observation. The instructions correctly require user confirmation for registry selection which is good, but they also reference environment variables (SKILLHUB_API_TOKEN, CLAWHUB_API_TOKEN, CLAWHUB_TOKEN) and local filesystem access without that being declared in the manifest — a mismatch worth flagging.

Install Mechanism

There is no install spec (instruction-only), which is lower risk because no archives are downloaded or code written to disk by the skill bundle itself. However, the runtime relies on 'npx' to fetch/execute the 'clawhub' CLI at runtime; this reliance on npx/npm network fetches is not documented in the package metadata and should be disclosed to users.

Credentials

The skill declares no required environment variables, but the instructions tell the agent to check for and possibly use SKILLHUB_API_TOKEN, CLAWHUB_API_TOKEN, and CLAWHUB_TOKEN, or to ask the user for an API token to log in. Requesting an API token is expected for publishing, but the manifest should declare that tokens/credentials are used. Because tokens allow publish actions and could be sensitive, the absence of explicit env var requirements and lack of guidance about token scope or lifetime is a proportionality/clarity concern.

Persistence & Privilege

The skill does not request persistent or always-on privileges (always:false) and does not attempt to modify other skills or system-wide settings. Autonomous invocation is allowed (platform default) but not combined here with other high-privilege requests, so no additional persistence-related concern is evident.

Scan Findings in Context

[no_regex_findings] expected: The package is instruction-only and contained no code files for the regex scanner to analyze. That explains the lack of findings but does not guarantee safety; behavioral risks are expressed in the SKILL.md instructions.

What to consider before installing

This skill appears to do what it says (use the clawhub CLI to inspect and publish skills) but its manifest is incomplete. Before installing or letting an agent run it: (1) ensure 'npx' / Node/npm will be available where the agent runs — the SKILL.md assumes it but the package metadata does not declare it; (2) be cautious about providing API tokens: prefer short-lived, least-privilege tokens and do not share long-lived credentials; (3) when asked to publish a local folder, verify the folder contents yourself (avoid including secrets, private keys, or credentials in the published package); (4) consider running the 'inspect' and 'search' commands yourself first to validate the registry and the clawhub CLI behavior; and (5) ask the skill author to update the metadata to explicitly declare required binaries and environment variables (npx, SKILLHUB_API_TOKEN/CLAWHUB_*), or treat the tool as 'read-only' (search/inspect) unless you explicitly consent to publishing with a controlled token. If you want, I can list the exact commands the agent would run and which pieces of data it will request so you can decide what to allow.

This skill gives you the ability to interact with a SkillHub ecosystem using the clawhub CLI. It supports previewing available skills and uploading/publishing new skills securely to the server.

Overview

SkillHub instances host agents, profiles, and skills securely. As an AI Agent, you might be asked to publish a local folder as a skill, or to preview an existing skill to see what it does.

No need to write raw HTTP REST requests; SkillHub provides full compatibility with the npx clawhub toolchain.

Before using this skill for any live action, you must ask the user for the SkillHub address they want to use. Do not assume the default registry unless the user explicitly confirms it.

If the user has not provided a SkillHub address yet, ask for it first. After the user gives the address, use that address consistently for login, search, inspect, and publish commands.

Mandatory Interaction Pattern

Before any registry action, follow this exact sequence:

1. Ask: Please provide the SkillHub address you want me to use.
2. Wait for the user's reply.
3. Repeat the address back to the user in a confirmation message.
4. State whether you will use plain npx clawhub ... or CLAWHUB_REGISTRY=<address> npx clawhub ....
5. Only after that confirmation message may you run login, search, inspect, explore, or publish.

Example confirmation:

• Confirmed. I will use SkillHub at https://your-registry.example and run subsequent clawhub commands with CLAWHUB_REGISTRY set to that address.

If the user says to use the default hosted registry, confirm that explicitly before proceeding.

Never skip the repeat-back confirmation step.

General Authentication & Environment

All SkillHub actions must point to the intended registry.

You must explicitly ask the user which SkillHub address to use before running any registry command.

Once the user provides the address:

• if it is the default hosted registry, you may use plain npx clawhub <command>
• if it is a custom or self-hosted registry, set CLAWHUB_REGISTRY=<user-provided-url> for every command

Examples:

• default registry: npx clawhub <command>
• custom registry: CLAWHUB_REGISTRY=https://your-registry.example npx clawhub <command>

Never silently pick a registry on behalf of the user.

Is Login Required?

• Viewing public skills: usually no login required.
• Publishing or interacting with private/team spaces: login required.

If you need to login before publishing:

• explicit token: npx clawhub login --token <TOKEN>
• custom registry plus token: CLAWHUB_REGISTRY=https://your-registry.example npx clawhub login --token <TOKEN>
• if the environment already provides SKILLHUB_API_TOKEN, CLAWHUB_API_TOKEN, or CLAWHUB_TOKEN, validate first with npx clawhub whoami
• if no working token is available, ask the user for one before proceeding

Commands

See references/workflows.md for full syntax and step-by-step examples of:

1. Publishing a skill (npx clawhub publish)
2. Previewing and Inspecting a skill (npx clawhub inspect)
3. Searching for skills (npx clawhub search)