Security audit
Security checks for vulnerabilities and agentic risk
This is a read-only Intercom conversations skill with expected token use; the main cautions are sensitive conversation data handling and dependency hygiene, not evidence of malicious behavior.
Before installing, use a least-privilege Intercom token and be careful about what conversation data you ask the agent to retrieve, because customer support content may become part of the agent context. Prefer installing with a lockfile or exact vetted dependency version for reproducibility.
Code accesses credential files (SSH keys, AWS credentials, etc.). This could indicate credential theft attempts.
{
"name": "INTERCOM_ACCESS_TOKEN",
"required": true,
"description": "Intercom Personal Access Token / OAuth access token."
}
],
"actions": [
Code accesses credential files (SSH keys, AWS credentials, etc.). This could indicate credential theft attempts.
{
"name": "INTERCOM_ACCESS_TOKEN",
"required": true,
"description": "Intercom Personal Access Token / OAuth access token."
}
],
"actions": [
The documentation states that an Intercom access token is required and describes actions that list, find, and search conversations, but it does not explicitly warn users that the skill will transmit that token and potentially sensitive conversation data to the external Intercom API. Because conversation content may include customer support messages and other sensitive business data, lack of clear disclosure can lead to uninformed use and inadvertent data exposure outside the local agent environment.
The skill description says it is invoked as default(input), which indicates a broad default trigger without any documented activation constraints, user-consent gates, or narrowing conditions. In a read-capable Intercom conversations skill, this increases the chance that an agent may call it opportunistically and retrieve sensitive customer support conversation data without sufficiently explicit user intent.
The dependency is specified with a caret range (^6.0.0), which allows newer 6.x releases to be installed without review. In a security-sensitive integration skill that reads customer conversations from Intercom, this increases supply-chain risk because a later compromised or vulnerable minor/patch release could be pulled in during installation.
"node": ">=18"
},
"dependencies": {
"intercom-client": "^6.0.0"
},
"license": "MIT",
"private": true
The manifest references intercom-client without pinning to a specific vetted release, while the package has known advisories including reports of compromised/malicious published versions. Because this skill accesses Intercom conversation data, installing an affected version could expose sensitive customer communications or execute attacker-controlled code in the skill environment.
No suspicious patterns detected.