The Agent Skills Directory

[SAFE]: The skill provides a modular and robust SDK for interacting with Tableau's REST and Metadata (GraphQL) APIs. No malicious code, obfuscation, or unauthorized network operations were detected.
[COMMAND_EXECUTION]: The skill utilizes a 'CodeAct' approach, instructing the agent to execute Python code directly via uv run python -c for site exploration and data querying. This is the intended operational model for the skill and is well-documented.
[CREDENTIALS_UNSAFE]: The skill follows secure practices by advising users to store sensitive credentials (PATs, passwords) in environment variables or .env files. The provided data.py module includes a defensive scrubbing mechanism to ensure that credentials are never accidentally persisted to the temp/ directory.
[PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection because it ingests data from external Tableau environments that may be controlled by third parties.
Ingestion points: Untrusted data enters the agent context through schema introspection (Session.introspect) and VDS query execution (Session.query) in src/query_tableau_data_py/session.py.
Boundary markers: The instructions do not specify the use of clear delimiters or 'ignore' instructions for the data payloads retrieved from Tableau.
Capability inventory: The skill has the capability to execute shell commands via uv run, write files to the temp/ directory via src/query_tableau_data_py/data.py, and make network requests to the configured Tableau server.
Sanitization: There is no evidence of sanitization or filtering of external metadata (e.g., field descriptions, project names) before it is surfaced to the agent's reasoning context.

query-tableau-data