The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill requires the installation of a global Bun package @kyoji2/intercom-cli. This is a vendor-provided resource necessary for the skill's primary functionality.
[COMMAND_EXECUTION]: The agent is instructed to execute shell commands using the intercom CLI to perform actions such as searching contacts, replying to conversations, and managing articles.
[PROMPT_INJECTION]: Indirect Prompt Injection Risk.
Ingestion points: The skill retrieves untrusted data (customer messages, internal notes, and help center articles) via commands such as intercom conversation get, intercom conversation search, intercom article search, and intercom contact notes.
Boundary markers: There are no explicit delimiters or instructions provided in the skill files to help the agent distinguish between administrative instructions and content retrieved from external sources.
Capability inventory: The agent has the ability to execute a wide range of CLI commands (intercom ...) which could be misused if the agent follows malicious instructions embedded in conversation bodies or notes.
Sanitization: There is no evidence of content sanitization or validation of retrieved data before it is presented to the agent's context.
[CREDENTIALS_UNSAFE]: The skill manages sensitive Intercom Access Tokens. These are stored in a local configuration file at ~/.config/intercom-cli/config.json. Additionally, providing tokens via command-line arguments (e.g., intercom login "token") may cause sensitive data to be recorded in shell history files.

intercom