An agent skill for implementing, debugging, reviewing, refactoring, and migrating code
where correctness matters. It asks for the smallest complete solution, proof that matches
the risk, and an honest COMPLETE / PARTIAL / BLOCKED report.
- Three modes. Focused, Substantial, and High-risk. The mode sets how much process applies, so a small fix doesn't get migration-level ceremony. High-risk triggers (auth, payments, migrations, concurrency, public APIs, and others) are listed explicitly.
- One loop. Inspect, pin the contract and invariant, get a baseline, change the smallest coherent slice, then verify. A failure is sorted as an implementation defect, a bad assumption, an environment problem, or an ambiguous contract, and the agent loops back to the matching step.
- Delegation, with one writer. Fan out only for independent, read-only slices; brief each worker with an objective, scope, output contract, budget, and timeout; treat returns as evidence, not verdicts; report failed or unrun slices as gaps; review independently and keep synthesis in the parent.
- Evidence scaled to risk. Pre-fix evidence, recorded commands, and independent review scale with the mode. A skipped or unavailable check never counts as a pass.
- Design rules. Seven rule groups for the code the agent writes (ownership, boundaries, effects, failure, dependencies, change surface, tests) with a priority order for when they conflict, and review questions to answer against non-trivial diffs.
- References on demand. Quality gates for affected domains and a report template for non-trivial work.
Copy or symlink this directory into your agent's skills folder as code-max, for example:
git clone https://github.com/PyModel/code-max.git ln -s "$PWD/code-max" ~/.claude/skills/code-max
Other hosts use their own skills paths: ~/.agents/skills, ~/.cursor/skills,
~/.gemini/skills, ~/.pi/agent/skills, or ~/.config/opencode/skills.
Use code-max to fix the token refresh race. Preserve existing APIs.
Use code-max for a read-only review of the payment retry path.
The skill says to skip trivial edits.
Instructions can't force an agent to comply or sandbox what it runs, so you still need host permissions, code review, and CI. The skill grants no authority to push, deploy, or override project rules.