Skip to content
Projects

memcpython/no-slop has 16 GitHub stars and 2 forks. GitHot measured 0 stars in 24 hours. Last checked 2026-11-15 04:44 UTC. Last push 2026-11-11.

How these counts are measured

Measured star growth

in 24h
0
in 7 days
Collecting history
in 15 days
Collecting history
in 1 month
Collecting history

Net star changes over overlapping periods, not daily totals. Missing measurements are left blank. Stars measure attention, not software quality.

About

Make AI coding agents think before coding. No slop, no overengineering, no slop.

AI

README.md

Original

NO SLOP. An animated four-step workflow: read the code, trace the root cause, make the right change, and verify the result.

no-slop

Stop shipping code that only looks correct.

The point · The rules · Install · Read the skill


The point

AI can write a hundred lines before understanding the first one.

It can invent APIs, bury bugs under fallback logic, rewrite working code because reading it takes effort, and then tell you the implementation is production-ready.

no-slop pushes back against that.

One SKILL.md. No runtime. No dependencies. Just a set of rules for agents that write, debug, refactor, or review code.

Think first. Change less. Prove what works.

The rules

No more Do this instead
Guessing how an API works Check the installed version and actual contract
Patching symptoms Trace the failure to its cause
Rewriting working code for convenience Make the smallest coherent change
try/catch everywhere Handle errors where recovery makes sense
Comments that narrate the obvious Explain only what the code cannot
Random managers, wrappers, and factories Add abstractions only when they earn their keep
“Fixed” without a test or reproduction Say exactly what was verified

The skill also requires checking callers, preserving behavior, distinguishing evidence from assumptions, and reporting uncertainty instead of bluffing.

Install

Download SKILL.md and place it in your agent's skills directory:

Agent Project-local User-wide
Codex .agents/skills/no-slop/SKILL.md ~/.agents/skills/no-slop/SKILL.md
Claude Code .claude/skills/no-slop/SKILL.md ~/.claude/skills/no-slop/SKILL.md

Or install it into any agent that supports the Agent Skills format.

Start a new agent session after installing. If your agent doesn't load the skill automatically, invoke no-slop explicitly in your coding request.

Example: what changes in practice?

Instead of:

def get_name(user):
    try:
        return user.name
    except Exception:
        return "unknown"

The agent should first establish whether user.name is guaranteed by the actual contract. If it is:

def get_name(user):
    return user.name

If missing names are valid, handle that specific case. Swallowing every exception isn't robustness; it hides failures.

The real task is understanding the contract, not making the code shorter.

What this is / isn't

It is: an instruction-only skill that encourages careful investigation, focused changes, useful comments, and evidence-backed verification.

It isn't: a linter, a security scanner, a code-quality benchmark, or a guarantee that a model will follow every instruction. Results depend on the agent, project, and task.

The entire skill lives in SKILL.md. Read it, fork it, change what you disagree with.


MIT licensed. See LICENSE.