analysis-technical

MEDIUM W011: Third-party content exposure detected (indirect prompt injection risk).

• Third-party content exposure detected (high risk: 0.85). Outsider free text can enter the LLM context via the runtime-provided OHLCV JSON file/STDIN (--input path or -), whose history/data rows are arbitrary text from outside the agent; ta_compute.py reads it with Path(...).read_text() / sys.stdin.read() and parses it into JSON used for computation.