A pre-install verification gate that formally proves an AI agent skill cannot exceed its declared capabilities before allowing it onto your system

26.1% of agent skills across major registries have at least one security vulnerability according to a 42,447-skill empirical study. Snyk found 13.4% of ClawHub skills contain critical issues. Current scanners use pattern matching and heuristics, which miss novel attack vectors. This tool uses formal verification to mathematically prove that a skill's actual behavior matches its declared capability set, blocking installation if the proof fails. It sits as a pre-install gate in the OpenClaw skill lifecycle.

Demand Breakdown

405

Social Proof 2 sources

From SKILL.md to Shell Access in Three Lines of Markdown

2026-02-20

225 HN

One command turns any open-source repo into an AI agent backdoor

2026-03-15

180

Gap Assessment

CompetitiveMultiple tools exist but differentiation opportunities remain

3 tools exist (SkillFortify, Snyk Agent Scan, Cisco Skill Scanner) but gaps remain: Only 20 GitHub stars, very early stage. No pre-install gate integration with OpenClaw's skill lifecycle. Academic proof-of-concept, not production-ready CLI.; Heuristic-based, not formal verification. Cannot mathematically prove a skill stays within bounds. Pattern matching misses novel attack vectors that formal analysis would catch..

Features3 agent-ready prompts

Capability extractor that parses SKILL.md declarations and builds a formal specification of what the skill claims it can do

▶

Static analyzer that traces all code paths in the skill source and builds a behavioral model of what it actually does at runtime

▶

Proof engine that compares declared capabilities against observed behavior and outputs pass/fail with specific violation evidence

▶

Competitive LandscapeFREE

Product	Does	Missing
SkillFortify	Formal security scanner for 22 agent frameworks with static analysis and SBOM generation	Only 20 GitHub stars, very early stage. No pre-install gate integration with OpenClaw's skill lifecycle. Academic proof-of-concept, not production-ready CLI.
Snyk Agent Scan	Detects 15+ security risks across MCP servers and agent skills including prompt injection and tool poisoning. 2.4K stars.	Heuristic-based, not formal verification. Cannot mathematically prove a skill stays within bounds. Pattern matching misses novel attack vectors that formal analysis would catch.
Cisco Skill Scanner	IDE-integrated security scanner for agent skills with 1.9K stars. Brings security visibility to the development toolchain.	IDE-focused, not a pre-install gate. Scans during development, not at install time. No formal verification, uses rule-based detection.

Aggregate Score

12,965

0 leads found

Details

TypeProduct Idea

Competitors3

Features3

Issues2

Leads0

Source Signals

All signals →

7.9KClawHavoc Campaign: 824+ Malicious ClawHub Skills, 12% of Marketplace Is Malware 4.8KSnyk ToxicSkills: 36% of ClawHub Skills Have Prompt Injection, 1,467 Malicious Payloads 22526.1% of 42,447 Agent Skills Have Security Vulnerabilities — Large-Scale Empirical Study 0SkillSieve: Three-Layer Triage Detects Malicious AI Agent Skills at 0.800 F1 for $0.006/Skill

Related Ideas

All ideas →

0A CLI tool that scans a running OpenClaw instance for every known CVE, exposed endpoint, malicious skill, and token scope violation, then outputs a prioritized remediation checklist 0A runtime behavioral sandbox that detects guidance injection attacks in OpenClaw skills by observing what agents actually do instead of scanning what skills say 0A CLI tool that scans a running OpenClaw instance, scores its security posture, maps plugin dependencies to alternative platforms, and outputs a stay-or-migrate recommendation with effort estimates