Skillv1.0.0

ClawScan security

Lead Enricher - Explorium AgentSource · ClawHub's context-aware review of the artifact, metadata, and declared behavior.

Scanner verdict

ReviewMar 1, 2026, 6:56 AM

Verdict: Review
Confidence: high
Model: gpt-5-mini
Summary: The skill appears to implement the stated B2B lead-enrichment functionality and only needs an Explorium API key, but the package metadata and runtime instructions are inconsistent (the registry claims no credentials while SKILL.md/plugin.json and the CLI require EXPLORIUM_API_KEY), which is a notable coherence issue you should resolve before installing.
Guidance: This package mostly does what it says: it installs a local CLI that talks only to api.explorium.ai and needs your Explorium API key. Before installing: 1) Confirm the source/author (the registry metadata you were shown omitted the API-key requirement — that mismatch could be a packaging mistake). 2) Inspect bin/agentsource.py and setup.sh locally (they are included) and run setup.sh yourself only if you trust the code. 3) Keep your EXPLORIUM_API_KEY secret: set it as an environment variable or save it to ~/.agentsource/config.json (mode 600) rather than pasting into chat. 4) Consider running the install in an isolated account/container if you don't fully trust the publisher; verify that the CLI only contacts api.explorium.ai and that --call-reasoning is used only with explicit consent. 5) If you plan to install permanently, ask the publisher/maintainer to correct the registry metadata so required credentials are declared consistently.

Review Dimensions

Purpose & Capability: noteThe declared purpose (search/enrich companies and prospects via Explorium AgentSource) matches the included CLI and documentation: network calls target https://api.explorium.ai/v1/ and the code provides autocomplete/fetch/enrich/export flows. However the registry-level 'Requirements' block (shown earlier) claims no required env var/primary credential, while SKILL.md and plugin.json clearly require EXPLORIUM_API_KEY. That mismatch is an inconsistency in packaging/metadata.
Instruction Scope: okSKILL.md instructs the agent to locate and run the provided CLI, read its temp-file outputs, and optionally include the user's query as --call-reasoning only with consent. The runtime steps (checking ~/.agentsource/config.json, reading /tmp/agentsource_*.json, calling the Explorium API) are within the stated purpose. The skill explicitly disallows pasting API keys into chat and documents privacy concerns for call_reasoning.
Install Mechanism: okThere is no remote download: setup.sh copies the included bin/agentsource.py into ~/.agentsource/bin and optionally writes a local config.json (mode 600). The installer does not fetch code from arbitrary URLs and the CLI is pure-stdlib Python. This is low-risk from an install-fetch perspective, but it does create files under the user's home directory (/home/.../.agentsource) and uses /tmp for output.
Credentials: concernFunctionality requires a single service credential (EXPLORIUM_API_KEY) which is proportionate for a third-party API client. The concern is the manifest inconsistency: the top-level registry requirements reported 'none', while plugin.json and SKILL.md require and document EXPLORIUM_API_KEY and local storage at ~/.agentsource/config.json. Verify which metadata the platform will honor before installing. No other unrelated credentials or secrets are requested in code or docs.
Persistence & Privilege: okalways:false (normal). The skill installs files to ~/.agentsource and may save an API key to ~/.agentsource/config.json (mode 600) if the user opts in — this is expected for a CLI tool. It does not request system-wide privileges or modify other skills. Note: temp results are written to /tmp and could be readable by other local users/processes depending on OS temp-file policies.