Security audit
Security checks for vulnerabilities and agentic risk
This skill is a straightforward Exa web-search wrapper, but users should remember that searches and submitted URLs go to Exa's external API.
Install this only if you are comfortable using Exa as a third-party search provider. Do not submit secrets, private repository details, internal-only URLs, regulated data, or confidential research targets unless you are authorized to send them to Exa.
The skill exposes shell-based execution paths via documented bash commands, but does not declare permissions or clearly scope those capabilities. That mismatch can cause users or orchestration systems to treat the skill as less privileged than it really is, increasing the chance of unintended command execution or unsafe trust decisions.
The documentation describes sending search queries and URL content to Exa, but does not explicitly warn that user-provided queries and URLs will be transmitted to an external third-party service. This creates a data leakage risk because users may supply sensitive terms, internal URLs, or confidential research targets without realizing that the information leaves the local environment.
The script sends the user-supplied query to Exa's external API, but it does so silently with no user-facing notice, confirmation, or warning about network transmission. In an agent skill context, queries may contain sensitive internal code, secrets, URLs, or proprietary research terms, so undisclosed exfiltration to a third party creates a real confidentiality risk.
The script sends user-supplied URLs to a third-party service for remote content extraction, which can expose sensitive targets, internal URLs, or user browsing/research intent without any explicit warning or confirmation. In an agent skill context this is more dangerous because users may assume processing is local, while the API call causes data disclosure to an external provider.
The hardcoded endpoint https://api.exa.ai/search confirms that the skill is designed to transmit data off-box to a remote vendor-controlled service. That is not inherently malicious, but in this context it increases risk because the skill description encourages searching code, research, people, and companies—categories that often include confidential or regulated information.
highlights: true
}')
curl -s -X POST 'https://api.exa.ai/search' \
-H "x-api-key: $EXA_API_KEY" \
-H 'Content-Type: application/json' \
-d "$PAYLOAD"
The hardcoded endpoint https://api.exa.ai/search confirms that the skill is designed to transmit data off-box to a remote vendor-controlled service. That is not inherently malicious, but in this context it increases risk because the skill description encourages searching code, research, people, and companies—categories that often include confidential or regulated information.
highlights: true
}')
curl -s -X POST 'https://api.exa.ai/search' \
-H "x-api-key: $EXA_API_KEY" \
-H 'Content-Type: application/json' \
-d "$PAYLOAD"
The hardcoded Exa API endpoint confirms that the script is designed to send data off-host to a remote service. While not malicious by itself, this increases exposure because all requested URL-processing is delegated externally, which can leak sensitive inputs and metadata in a security-sensitive agent environment.
}')
# Call API
curl -s -X POST 'https://api.exa.ai/contents' \
-H "x-api-key: $EXA_API_KEY" \
-H 'Content-Type: application/json' \
-d "$PAYLOAD"
The hardcoded Exa API endpoint confirms that the script is designed to send data off-host to a remote service. While not malicious by itself, this increases exposure because all requested URL-processing is delegated externally, which can leak sensitive inputs and metadata in a security-sensitive agent environment.
}')
# Call API
curl -s -X POST 'https://api.exa.ai/contents' \
-H "x-api-key: $EXA_API_KEY" \
-H 'Content-Type: application/json' \
-d "$PAYLOAD"
No suspicious patterns detected.