The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill recommends the installation and use of several third-party Node.js packages and remote MCP servers to extend agent capabilities. These include packages such as @modelcontextprotocol/server-sequential-thinking, @modelcontextprotocol/server-memory, tavily-mcp, and @21st-dev/magic, as well as remote HTTP services like mcp.context7.com and stitch.googleapis.com.
[DATA_EXFILTRATION]: The configuration wizard provides an optional telemetry feature (Step 10) that allows users to stream session events, analytics, and summaries to an external webhook URL. While user-configured and transparently documented, this represents a mechanism for data exfiltration from the agent session to external infrastructure.
[COMMAND_EXECUTION]: The skill performs file system operations using the Bash and Write tools to save configuration settings to ~/.claude/plugins/orchestkit/config.json. It also provides instructions for managing environment variables and shell profile modifications to persist settings.
[INDIRECT_PROMPT_INJECTION]: As a configuration tool that ingests user-supplied arguments and interactive inputs to generate shell commands and JSON configuration files, the skill presents a surface for indirect prompt injection. The skill includes specific validation and backup rules (rules/backup-before-modify.md and rules/validation-after-write.md) to mitigate risks associated with malformed or malicious configuration data.

configure