The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill downloads the endorctl binary from the official vendor domain api.endorlabs.com. This is a standard and necessary step for the skill's primary function of setting up the Endor Labs CLI tool.
[REMOTE_CODE_EXECUTION]: After downloading the tool, the skill grants execution permissions and runs the binary. It includes a security best practice by verifying the binary's integrity against a SHA-256 checksum fetched from the same trusted vendor API.
[COMMAND_EXECUTION]: The skill executes shell commands to perform environment setup tasks, such as creating a local binary directory (~/bin), moving files, and updating the shell's PATH variable. These operations are routine for CLI tool installations.
[PROMPT_INJECTION]: The skill features an automated workflow that fetches command-line flags from the vendor's documentation site (docs.endorlabs.com) and interpolates them into execution commands. This constitutes an indirect prompt injection surface where the agent's actions could be influenced by external web content.
Ingestion points: The skill uses WebFetch to read content from https://docs.endorlabs.com/endorctl/commands/scan/ inside the SKILL.md workflow.
Boundary markers: No explicit delimiters or instructions are used to separate the fetched content from the rest of the command string.
Capability inventory: The skill has the capability to execute shell commands via subprocesses with the fetched parameters.
Sanitization: No explicit validation or filtering of the flags extracted from the documentation is performed before they are used in commands.

endor-setup