The Agent Skills Directory

[DATA_EXFILTRATION]: The skill instructs the agent to print sensitive environment variables, specifically SCALEKIT_CLIENT_SECRET, to the terminal using echo. This practice exposes credentials in session logs and terminal history. Furthermore, it transmits these credentials to an external endpoint ($SCALEKIT_ENVIRONMENT_URL/oauth/token) using curl. While these operations are directed at the vendor's legitimate infrastructure for validation purposes, they involve the exposure and transmission of sensitive authentication data.
[COMMAND_EXECUTION]: The skill executes shell commands, including curl for network requests and rg (ripgrep) for searching the local filesystem. These commands are used to verify the environment and search for hardcoded secrets.
[INDIRECT_PROMPT_INJECTION]: The skill ingests data from the local repository through a filesystem search (rg) in Step 2. If the files being scanned contain adversarial instructions, they could influence the agent's behavior during the go-live check.
Ingestion points: Local source code files scanned via the rg command in SKILL.md.
Boundary markers: None; there are no instructions to the agent to ignore or delimit embedded content found during the search.
Capability inventory: The agent has access to environment variables, filesystem visibility, and the ability to perform network requests via curl.
Sanitization: No sanitization or validation of the content found by the search command is performed before the agent evaluates it against the go-live checklist.

check-agentkit-prod