The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and process large volumes of untrusted external financial data, including news headlines, management guidance text, and corporate governance records.\n
Ingestion points: Data enters the agent's context through lseg.data library calls and browser-based extraction via scripts/workspace_cdp.py and scripts/ws_fetch.py.\n
Boundary markers: The SKILL.md file defines an "IRON LAW" for "Query Enforcement," instructing the agent to validate and inspect data samples (e.g., using .head() or .sample()) before presenting it to the user. This serves as a procedural boundary.\n
Capability inventory: The agent has the capability to execute Python scripts, perform network requests to LSEG domains, and write data to the local filesystem (e.g., in scripts/guidance_pull.py).\n
Sanitization: There is no evidence of programmatic sanitization or filtering of the retrieved financial text to prevent embedded instructions from influencing the agent's behavior.\n- [DYNAMIC_EXECUTION]: Several scripts in the skill utilize the Chrome DevTools Protocol (CDP) to execute JavaScript dynamically within a browser session.\n
Evidence: scripts/workspace_cdp.py, scripts/ws_fetch.py, and scripts/dib_query.py use the Runtime.evaluate method to perform tasks such as lifting authentication tokens from localStorage, interacting with the DOM (Data Item Browser), and executing fetch calls to internal LSEG endpoints.\n
Pattern: Executable JavaScript is constructed from strings and executed at runtime, which is a standard pattern for browser automation but constitutes dynamic code execution.\n- [COMMAND_EXECUTION]: The skill includes multiple Python scripts meant for execution by the agent to perform connectivity tests and batch data retrieval.\n
Evidence: scripts/test_connection.py, scripts/guidance_pull.py, and scripts/dib_query.py are provided as functional tools within the skill.\n- [EXTERNAL_DOWNLOADS]: The skill documentation references the installation of external dependencies from public registries.\n
Evidence: Instructions mention installing lseg-data, pandas, and websocket-client via pip. These are well-known, reputable packages used for financial analysis and WebSocket communication.

lseg-data