Skip to content

QODER.md

Latest commit

 

History

History
110 lines (78 loc) · 4.21 KB

QODER.md

File metadata and controls

110 lines (78 loc) · 4.21 KB
Raw

QODER.md

Behavioral guidelines to reduce common LLM coding mistakes when using Qoder in PyCharm. 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.
  • Prefer PyCharm's built-in refactoring tools (Rename, Extract, Move, etc.) over manual text manipulation when the IDE can do it safely.

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 happen before implementation rather than after mistakes.

Project-Specific Guidelines — jolarca-data

This repository governs pseudonymous analytics over marketplace data; every merge either changes the governance record or transforms data that was personal. The full rules live in CONTRIBUTING.md; these constrain AI-assisted changes:

Personal data boundaries

  • Never generate, fetch, or commit personal data — real or plausibly real. Fixtures come from synthetic/generators/ and are faker-seeded.
  • Never re-introduce identifiers into warehouse models: staging hashes IDs and drops names/emails; any model that joins them back is a critical defect.
  • Compliance marts stay aggregates-only; never emit per-subject rows.

Pseudonymization & retention

  • Changes to ingestion/pipelines/pseudonymizer/ or lifecycle/ are critical-risk: require DPO review and make anonymize-verify evidence before merge.
  • Retention jobs execute the schedule defined in governance/retention-map.md; policy text lives in jolarca-compliance — do not duplicate or reword policy here.
  • Legal holds suspend retention; they never delete.

Governance gates

  • New datasets must land with catalog entry + ownership row + classification + retention class; scripts/catalog-lint.py enforces.
  • make check (seed schema + catalog lint + PII tripwire) must pass before proposing a change as complete.
  • Never suggest bypassing pre-commit (--no-verify) or loosening the PII scan patterns to make a commit pass.

Secrets & access

  • No warehouse credentials, connection strings, or API keys in any file; profiles.yml.example/.envrc.example document env vars only.
  • Extraction connects to the read replica with a read-only role — never propose production credentials for analytics (ADR-0002).