CONTRIBUTING.md

Thinking Guidelines

Behavioral guidelines to reduce common LLM coding mistakes. Merge with project-specific instructions as needed.

Tradeoff: These guidelines bias toward caution over speed. For trivial tasks, use judgment.

(1) Think Before Coding

Don't assume. Don't hide confusion. Surface tradeoffs.

Before implementing:

• State your assumptions explicitly. If uncertain, ask.
• If multiple interpretations exist, present them - don't pick silently.
• If a simpler approach exists, say so. Push back when warranted.
• If something is unclear, stop. Name what's confusing. Ask.

(2) Simplicity First

Minimum code that solves the problem. Nothing speculative.

• No features beyond what was asked.
• No abstractions for single-use code.
• No "flexibility" or "configurability" that wasn't requested.
• No error handling for impossible scenarios.
• If you write 200 lines and it could be 50, rewrite it.

Ask yourself: "Would a senior engineer say this is overcomplicated?" If yes, simplify.

(3) Surgical Changes

Touch only what you must. Clean up only your own mess.

When editing existing code:

• Don't "improve" adjacent code, comments, or formatting.
• Don't refactor things that aren't broken.
• Match existing style, even if you'd do it differently.
• If you notice unrelated dead code, mention it - don't delete it.

When your changes create orphans:

• Remove imports/variables/functions that YOUR changes made unused.
• Don't remove pre-existing dead code unless asked.

The test: Every changed line should trace directly to the user's request.

(4) Goal-Driven Execution

Define success criteria. Loop until verified.

Transform tasks into verifiable goals:

• "Add validation" → "Write tests for invalid inputs, then make them pass"
• "Fix the bug" → "Write a test that reproduces it, then make it pass"
• "Refactor X" → "Ensure tests pass before and after"

For multi-step tasks, state a brief plan:

1. [Step] → verify: [check]
2. [Step] → verify: [check]
3. [Step] → verify: [check]

Strong success criteria let you loop independently. Weak criteria ("make it work") require constant clarification.

These guidelines are working if: fewer unnecessary changes in diffs, fewer rewrites due to overcomplication, and clarifying questions come before implementation rather than after mistakes.

Coding Guidelines

Simplicity:

• no code > simple code > clever code
• more dependencies is better than more code
• minimize indentation: guard first programming, early returns, happy path first
• maximize for code locality: a little duplication is okay. inline small functions (<5 LoC). no module-level globals or top-level helpers unless reused in multiple places
• avoid OOP primitives like classes and inheritance where possible
• use itertools and functools to fold deeply nested loops

Avoid LLM smells:

• no trailing commas. prefer single-line imports, dicts and lists over multi-line with trailing commas. take advantage of infinite line width
• avoid docstrings where possible. prefer tight technical comments starting with the pound # symbol
• avoid unnecessary print statements. do not add indentation in prints with whitespaces but with \t
• call code directly in if __name__ == "__main__": instead of defining a second main() function
• import statements must always be at the top of the file
• avoid from __future__ import annotations where possible

Robustness:

• for standalone scripts, always either use bash or UV scripts with inline dependencies
• use assert statements frequently as invariants. prefer crashing over failing silently. prefer assert over runtime exceptions
• maximize test coverage, without adding noise (except for standalone scripts)
• prefer typed enums over strings for states
• use match statements over multiple isinstance branches

Performance:

• use @cache where it makes sense
• have mechanistic sympathy, design with data oriented programming best-practices