The Agent Skills Directory

[COMMAND_EXECUTION]: The skill facilitates the registration of command-type MCP servers via the uip CLI tool, which enables the agent to wrap and execute local shell commands and subprocesses as MCP tools.
[EXTERNAL_DOWNLOADS]: The skill provides functionality to connect to external endpoints by allowing the agent to register remote and swagger server types that fetch data from user-specified HTTP URIs or OpenAPI specifications.
[DATA_EXFILTRATION]: The skill documentation includes instructions for managing Orchestrator resources and folder-scoped data. It mentions the AssetReferenceSubstitutor mechanism, which handles the runtime resolution of credential-based asset references stored in the Orchestrator platform.
[PROMPT_INJECTION]: The skill exposes a surface for indirect prompt injection by ingesting data from external discovery tools and using that content to generate tool metadata and schemas.
Ingestion points: Data returned from uip is resources describe and uip agenthub mcp-tools candidates CLI commands as described in references/is-activity-workflow.md.
Boundary markers: Absent; there are no instructions to use delimiters or ignore embedded instructions when processing CLI output.
Capability inventory: Full shell execution capabilities via the Bash tool and file system modification via Write and Edit tools.
Sanitization: Absent; the agent is instructed to compose metadata directly from the CLI's JSON responses without explicit validation or filtering logic mentioned in the prompt instructions.

uipath-mcp-servers