Perplexica Search

PassAudited by VirusTotal on May 11, 2026.

Overview

Type: OpenClaw Skill Name: perplexica-search-local Version: 1.0.3 The Perplexica Search skill is a well-implemented tool for local AI-powered research. The code in scripts/perplexica_search.py includes proactive security measures to prevent data exfiltration and SSRF, such as custom urllib handlers that explicitly reject HTTP redirects and a validation function that ensures resolved hostnames point only to loopback or private IP addresses (RFC 1918). No malicious intent, obfuscation, or unauthorized data access was found.

Findings (0)

Artifact-based informational review of SKILL.md, metadata, install specs, static scan signals, and capability signals. ClawScan does not execute the skill or run runtime probes.

What this means

You may need to trust the packaged artifact rather than a verified upstream source link.

Why it was flagged

The registry metadata does not identify a verified source even though a homepage is provided. This is a provenance note, not evidence of malicious behavior.

Skill content
Source: unknown; Homepage: https://github.com/eplt/perplexica-search
Recommendation

Review the included script and, if possible, compare it with the linked repository before installing.

What this means

Sensitive search terms or pasted conversation history will be processed by your Perplexica instance and whatever providers it is configured to use.

Why it was flagged

The script sends the user's query, and optionally custom instructions and conversation history, to the Perplexica API. This is expected for the skill's purpose and is restricted in the artifacts to local Perplexica URLs.

Skill content
"query": query, "stream": True ... payload["systemInstructions"] = instructions ... payload["history"] = history ... f"{base_url}/api/search"
Recommendation

Use a trusted local Perplexica instance and avoid sending sensitive history unless you are comfortable with that instance's provider configuration.

What this means

Search answers may reflect untrusted web content, and any retained search/RAG state is controlled by Perplexica rather than this skill.

Why it was flagged

The skill relies on Perplexica's retrieved context and RAG state. This is core to the search feature, but users should understand that retrieved web/source content can influence answers and state handling is delegated to Perplexica.

Skill content
returns answers with cited sources in OpenClaw while keeping search/RAG state in Perplexica
Recommendation

Verify important claims against cited sources and review Perplexica's own retention/configuration settings.

What this means

A user could overestimate privacy if their Perplexica instance uses remote search or model providers.

Why it was flagged

The local-only claim is supported for the script's direct target URL, but users should not interpret it as a guarantee that Perplexica's configured search or LLM providers are all local.

Skill content
Local-only; no data exfiltration ... At least one chat model provider configured in Perplexica
Recommendation

Check Perplexica's provider settings and treat the privacy claim as applying to this script's direct network target, not necessarily all downstream processing.