Skip to content

ai-ft.md

Latest commit

 

History

History
129 lines (99 loc) · 3.69 KB

ai-ft.md

File metadata and controls

129 lines (99 loc) · 3.69 KB

[AI Instruction Set: Absolute Compliance Required]

Overview
The AI assistant must fully comply with the following rules in every response. No exceptions. Any violation constitutes non-compliance.

1. Imports and Namespace Management

  • No from Imports
    Under no circumstances should the AI use from module import X.
  • Explicit Module References
    Always use import module and reference all attributes via module.attribute.

2. Documentation Standards

  • NumPy-Style Docstrings
    Every function, class, and module must include comprehensive NumPy-style docstrings with:
    • Summary
    • Parameters
    • Returns
    • Raises
    • Examples
    • Notes
  • Clarity
    Docstrings must clearly describe input types, output types, possible exceptions, and usage examples.

3. Code Completeness

  • Self-Contained
    Provide fully implemented, production-ready code. No placeholders or incomplete sections.
  • Ready to Execute
    All dependencies and logic must be included; the user should not need to add or modify anything to run the code.

4. Library Usage

  • Permitted
    Any open-source Python library that does not require network access.
  • Prohibited
    • HTTP or network-call libraries (requests, aiohttp, httpx, etc.).
    • External APIs or internet-based services.

5. Precision, Performance, and Security

  • Concise Responses
    Focus on clarity, correctness, and efficiency; avoid unnecessary exposition.
  • FinTech Best Practices
    Ensure code meets security, scalability, and data-integrity standards common in financial technology.
  • Modularity
    Design with extendibility in mind—components should be loosely coupled and easily testable.

6. Style and Type Safety

  • PEP Compliance
    Follow PEP 8 (style), PEP 20 (Zen of Python), PEP 257 (docstrings), PEP 484 (type hints).
  • Strict Typing
    Every function signature, variable, and return value must have accurate static type annotations.
  • No Type Warnings
    Code must pass strict type checking (e.g., MyPy, VS Code strict mode) without errors.

7. Operational Constraints

  • Local Execution Only
    No reliance on third-party web services or cloud infrastructure.
  • No Browser-Based Requests
    The solution must run entirely offline within the local environment.

8. Linting & Formatting Rules (Ruff Configuration)

Derived from pyproject.toml:

[tool.ruff]
extend-exclude = ["examples/*", ".venv/*"]
line-length = 79

[tool.ruff.lint]
select = [
  "E",    # Pycodestyle errors (PEP8)
  "F",    # Pyflakes (undefined names)
  "I",    # isort (import ordering)
  "TCH",  # Type-checking violations
  "C",    # Complexity (Cyclomatic, McCabe)
  "N",    # Naming conventions
  "D2",   # Pydocstyle: blank lines after summary
  "D3",   # Pydocstyle: multi-line docstring formatting
  "D415", # Sections end with colon
  "D417", # Docstring args match signature
  "D418", # Section capitalization
  "D419", # Final newline in docstring
  "ASYNC",# Async-specific linting
  "Q",    # Quotes consistency
  "RSE",  # Raise-statement correctness
  "SIM",  # Simplification rules
  "RUF",  # Ruff-specific rules
]
ignore = [
  "F405", # Allow wildcard imports internally
  "F403", # Allow from module import * internally
  "E501", # Disable strict line-length (handled above)
  "D205", # Disable one blank line after summary
  "D417", # Relax some arg-name mismatches
  "C901", # Temporarily allow high complexity
]
fixable = ["I", "TCH", "D"]

[tool.ruff.lint.pydocstyle]
convention = "numpy"

[tool.ruff.lint.isort]
force-single-line = true

[tool.ruff.format]
docstring-code-format = true