Back to skill
Skillv1.0.0
ClawScan security
lead-generation · ClawHub's context-aware review of the artifact, metadata, and declared behavior.
Scanner verdict
BenignApr 21, 2026, 1:30 AM
- Verdict
- benign
- Confidence
- medium
- Model
- gpt-5-mini
- Summary
- The skill's requirements and runtime instructions largely match its lead-generation purpose, but there are a few practical mismatches and privacy considerations you should review before installing.
- Guidance
- This skill appears to do what it claims (search social posts via SkillBoss) and only requests one API key. Before installing: (1) Verify you trust the SkillBoss service (api.skillbossai.com) and its handling of any social data you send; (2) Protect the SKILLBOSS_API_KEY (use a scoped key and rotate it if possible); (3) Confirm where the agent will write data (data/lead-generation/*.json) and whether that storage meets your privacy/compliance needs — these files may include people’s posts and identifiers; (4) Be mindful of outreach/legal/ethical rules for contacting individuals discovered this way and require user review before sending messages (the skill suggests drafting only); (5) If your agent provides web_fetch/web_search capabilities, check whether the skill will cause broader web access than advertised and whether that is acceptable. If you need higher assurance, request the maintainer’s privacy/security documentation and a concrete description of what SkillBoss does with submitted queries and returned data.
Review Dimensions
- Purpose & Capability
- okName/description (social lead discovery) align with the single required env var (SKILLBOSS_API_KEY) and with calls to the SkillBoss API Hub for searching social sites. The declared scope (Twitter/Instagram/Reddit search, scoring, outreach drafts) is coherent with the requested capability.
- Instruction Scope
- noteInstructions explicitly direct the agent to: call the SkillBoss API Hub, generate/iterate search queries, collect social posts/users, score leads, and write JSON files under data/lead-generation. They also mention using web_fetch/web_search for product research — which may access arbitrary web resources. The skill does not instruct reading unrelated system secrets, but it does create and persist local files that may contain collected user data; consider privacy and retention policy.
- Install Mechanism
- okThis is instruction-only with no install spec and no downloads. No code is written to disk by an install step, which reduces supply-chain risk.
- Credentials
- okOnly one credential (SKILLBOSS_API_KEY) is required, which fits the declared use of the SkillBoss API Hub. No unrelated secrets or multiple external credentials are requested.
- Persistence & Privilege
- noteThe skill will persist data to local files (data/lead-generation/*.json) for deduplication and profile storage. It does not declare system-wide or always-on privileges. If you are concerned about persistence, verify where the agent stores these files and how long data is retained/secured.