The Agent Skills Directory

[COMMAND_EXECUTION]: The skill uses the Bash tool to run tsh, tctl, terraform, and az commands. This is required for authenticating with Teleport, checking cluster status, and managing cloud resources.
[COMMAND_EXECUTION]: In SKILL.md, the agent is instructed to use eval "$(tctl terraform env)" to inject short-lived credentials into the shell environment for Terraform commands. This is a standard procedure for Teleport's Terraform integration.
[COMMAND_EXECUTION]: Supporting evaluation scripts in evals/run_evals.py and evals/_claude_util.py utilize the subprocess module to execute CLI tools and mock binaries during testing and grading.
[PROMPT_INJECTION]: The skill exhibits an indirect prompt injection surface as it processes structured output from tctl commands to diagnose enrollment failures.
Ingestion points: JSON output from tctl discovery nodes and tctl get user_tasks in SKILL.md.
Boundary markers: None. Output is parsed directly for failure details.
Capability inventory: Access to Bash (for Terraform) and file modification via Write/Edit tools.
Sanitization: No explicit sanitization or filtering of the tool output is described.
[SAFE]: The skill performs a WebFetch from https://goteleport.com/docs/... to retrieve troubleshooting guidelines. This targets the official documentation of the vendor and is considered a safe operation.

teleport-discovery