perplexity-web-mcp Security Audit Report

🔭 Continuously monitored by ClawSecure Watchtower

Source:

SHA-256:

perplexity-web-mcp is an AI agent skill, created by jacob-bd and published at jacob-bd/perplexity-web-mcp. ClawSecure audited perplexity-web-mcp across 72 files through the 3-Layer Audit Protocol covering all ten OWASP ASI Top 10 categories, assigning a security score of 50/100 (Warning). The 7 findings concentrate in Code Injection, ReDoS and Permissions Manifest, including Potentially dangerous code pattern detected: curl.*\|.*sh and Potentially dangerous code pattern detected: system\(. 3 were rated high or critical severity.

Is perplexity-web-mcp safe?

ClawSecure audited perplexity-web-mcp and assigned a security score of 50/100 (Warning), identifying 7 findings across Code Injection and ReDoS. Review the findings below before installing.

What did ClawSecure find in perplexity-web-mcp?

ClawSecure identified 7 findings in perplexity-web-mcp, concentrated in Code Injection, ReDoS and Permissions Manifest. 3 were rated high or critical severity. The most severe include Potentially dangerous code pattern detected: curl.*\|.*sh and Potentially dangerous code pattern detected: system\(.

How was perplexity-web-mcp audited?

ClawSecure ran perplexity-web-mcp through its 3-Layer Audit Protocol with full OWASP ASI Top 10 coverage, scanning 72 files from jacob-bd/perplexity-web-mcp.

What does a score of 50 mean?

ClawSecure assigned perplexity-web-mcp a security score of 50/100, placing it in the Warning range. This is driven by 7 findings led by Code Injection that should be addressed before use. ClawSecure derives this score with a weighted deduction model (critical -20, high -10, medium -5, low -2 from a base of 100).

Audit Findings for perplexity-web-mcp

ClawSecure detected 7 security findings in perplexity-web-mcp, spanning Code Injection, ReDoS and Permissions Manifest.

high · Potentially dangerous code pattern detected: curl.*\|.*sh. Code Injection finding detected in run_server.py.
high · Potentially dangerous code pattern detected: system\(. Code Injection finding detected in run_server.py.
high · Potentially dangerous code pattern detected: system\(. Code Injection finding detected in setup.py.
medium · Missing config.json - agent may not be properly configured. Permissions Manifest finding.
medium · ReDoS vulnerability: Nested quantifiers (potential…. ReDoS finding detected in SKILL.md:88.
medium · ReDoS vulnerability: Nested quantifiers (potential…. ReDoS finding detected in SKILL.md:88.
medium · ReDoS vulnerability: Nested quantifiers (potential…. ReDoS finding detected in shared.py:700.

Each finding is expandable in the interactive list below.

3-Layer Audit Protocol

Security Recommendations for perplexity-web-mcp

Eliminate dynamic code execution

perplexity-web-mcp evaluates code at runtime (for example eval or dynamic exec). Remove dynamic evaluation of untrusted input, and where code generation is unavoidable, sandbox it and validate every input.

Fix ReDoS-prone patterns

perplexity-web-mcp contains regular expressions vulnerable to catastrophic backtracking (ReDoS). Replace vulnerable patterns, bound input length, and prefer linear-time matching so a crafted input cannot hang the agent.

Add a config.json permissions manifest

A config.json file declares what an agent component can access: file system, network, shell execution and more. Without it, users have no visibility into what the component can do before installing. This is the single most impactful security improvement for any AI agent skill.

Related Security Research

Why Generic Scanners Fail at AI Agent Security→Beyond Static Scans: Why ClawSecure Verifies Agentic Intent→

Related AI Agent Security Audits

ai-skillsScore 35/100 ai-skillsScore 35/100 crewAIScore 65/100 anthropic-cookbookScore 55/100 remindctlScore 59/100

Scanned on September 1, 2026. perplexity-web-mcp is one of thousands of agents audited by ClawSecure from the community-curated awesome-openclaw-skills list and the openclaw/skills repository.

Start Your Free Scan