The Agent Skills Directory

[COMMAND_EXECUTION]: The installation instructions in instruct.md prompt the agent to append the QVERIS_API_KEY to the user's ~/.bashrc (or ~/.zshrc) file to ensure the credential persists across sessions. Modifying shell configuration files is a persistence mechanism.
[EXTERNAL_DOWNLOADS]: The instruct.md file contains instructions to fetch the skill definition (SKILL.md) and several implementation scripts (qveris_env.mjs, qveris_client.mjs, qveris_tool.mjs) from https://qveris.ai using curl.
[COMMAND_EXECUTION]: The skill documentation defines an invocation strategy (Tier 3) that relies on executing the node CLI tool to run local scripts with user-supplied arguments.
[PROMPT_INJECTION]: The skill processes data from an external API (capability discovery and tool output) which could contain malicious instructions designed to influence agent behavior (Indirect Prompt Injection surface).
Ingestion points: External data enters the context via the discover and call commands in qveris_tool.mjs.
Boundary markers: None present; output is displayed as formatted text or raw JSON.
Capability inventory: The environment supports shell command execution (node) and HTTP requests via fetch.
Sanitization: Descriptions are truncated for display, but full results are processed by the agent.

qveris-official