Add CONTRIBUTING: the bar, and why review here is unusual - #11
Merged
Conversation
This repo ships instructions that agents obey, so a one-line change to SKILL.md changes what every agent does next. States that review question explicitly, along with the bar a rule has to clear (its absence would let someone believe something false), the three tests for a new gauntlet layer, the in-one-out-one rule for keeping SKILL.md loadable, and the requirement that demo changes go through the loop the skill describes. Also names the failure mode this project keeps hitting -- mechanisms that report success while doing nothing -- and asks contributors to say what their change does when it is broken.
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Five PRs arrived from an outside fork this week, and reviewing them made it obvious the repo had never written down what it wants.
The parts that are specific to this project rather than boilerplate:
Review here is unusual.
skills/old-coder/is not documentation about code — it is instructions that agents obey. A one-line change toSKILL.mdchanges what every agent loading this skill does next. So the review question is not "is this sentence true" but "what will an agent do differently after this merges", and anything touching approvals, permissions, execution or writes outside the repo gets a security-shaped read.The bar. A rule earns its place if its absence would let someone believe something false — not if its presence would be nice to have. Same bar for EVIDENCE fields and gauntlet layers.
Three tests for a new layer: orthogonal, has been watched failing on a known-bad input, tool-ready. The second is where most proposals will stop, and that is intended.
In one, out one.
SKILL.mdis loaded in full on every invocation, including for tasks that never use the rule you are adding. A file long enough to be skimmed is its own fail-open.Demo changes go through the loop the skill describes, evidence rebound included. A demo change that skips the loop is not a demo of the loop.
It also asks contributors to say what their change does when it is broken, and points at the five times this project has shipped a mechanism that reported success while doing nothing.
🤖 Generated with Claude Code