search-first

Warn

Audited by Socket on Sep 19, 2026

4 alerts found:

Securityx3Anomaly
SecurityMEDIUM
SKILL.md

This skill is not a benign general research helper; it is an offensive-security workflow that teaches an AI agent to discover, adapt, and execute exploits while avoiding detection. There is no strong evidence of credential theft or hidden malware, but the capability set is clearly high risk and disproportionate for normal developer assistance.

Confidence: 92%Severity: 89%
SecurityMEDIUM
payloads.md

The visible code is a readable, dual-use reconnaissance and vulnerability-research toolkit. It contains no clear malware or covert exfiltration behavior, but it performs active scanning, exploit-source retrieval, credential-exposure searches, and outbound reporting. The eval-based quick_search function is a significant command-injection risk when inputs are untrusted, and API-key handling plus third-party data transmission require review. Do not use with untrusted arguments without removing eval and restricting network scope.

Confidence: 98%Severity: 72%
AnomalyLOW
guides/multi-source-intelligence-correlation-guide.md

The fragment is a legitimate vulnerability-intelligence correlation tool with no clear malicious behavior. It contains input-validation risks in the shell script: unsanitized TARGET values are used in filesystem paths and are directly interpolated into URL and GraphQL query text. These issues could enable path traversal, malformed or manipulated searches, and denial of service when invoked with untrusted input. Restrict TARGET to a safe identifier, URL-encode API parameters, construct GraphQL variables instead of string interpolation, and use a fixed or sanitized output directory.

Confidence: 98%Severity: 55%
SecurityMEDIUM
guides/exploit-research-methodology.md

The fragment is an offensive exploit-research and execution workflow. It contains explicit instructions and payloads for achieving Apache RCE, creating a substantial misuse and operational security risk, but it is not itself evidence of supply-chain malware or package compromise. Review and restrict execution in authorized, isolated environments only.

Confidence: 97%Severity: 78%
Audit Metadata
Analyzed At
Sep 19, 2026, 04:39 PM
Package URL
pkg:socket/skills-sh/brucesongs%2Fkali-claw%2Fsearch-first%2F@9331ad6f410731ff3ff8027ae621cbd72ef04f26c3269251361b15049517d382