You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
PR #17 audited every "Good" code example in powershell.instructions.md against the file's own rules and found multiple internal-rule violations: non-descriptive variable names, unquoted string literal parameters, an unapproved verb, missing parameter validation, etc. Each Good example in an instruction module is itself authoritative — agents pattern-match on examples, so a violation in one teaches the wrong thing.
The same audit hasn't been done for the other instruction modules in this repo.
Request
Audit each remaining *.instructions.md file under instructions/ and instruction-templates/ for "Good" examples that violate the file's own rules (or the rules of any other applicable instruction module — e.g., shorthand, markdown):
markdown.instructions.md — check fenced-block examples for compliance with their own listed rules
github-cli.instructions.md — check shell snippets for parameter quoting, full flag names, etc.
shorthand.instructions.md — check that examples don't themselves use shorthand
agent-workflow.instructions.md, readme.instructions.md, releases.instructions.md, repository-specific.instructions.md, update.instructions.md, contributing.instructions.md — check for any code examples and validate
Why a separate issue
The PR #17 audit was already substantial; expanding to other modules would have ballooned the scope. Separate PR keeps blame and review tractable.
Acceptance criteria
Each Good example in each instruction module passes every applicable rule (the file's own + cross-cutting modules like shorthand)
Both instruction-templates/ and instructions/ mirror copies updated together (no sync drift introduced)
CHANGELOG entry under [Unreleased] describing the audit and listing changes per module
Context
PR #17 audited every "Good" code example in
powershell.instructions.mdagainst the file's own rules and found multiple internal-rule violations: non-descriptive variable names, unquoted string literal parameters, an unapproved verb, missing parameter validation, etc. Each Good example in an instruction module is itself authoritative — agents pattern-match on examples, so a violation in one teaches the wrong thing.The same audit hasn't been done for the other instruction modules in this repo.
Request
Audit each remaining
*.instructions.mdfile underinstructions/andinstruction-templates/for "Good" examples that violate the file's own rules (or the rules of any other applicable instruction module — e.g., shorthand, markdown):markdown.instructions.md— check fenced-block examples for compliance with their own listed rulesgithub-cli.instructions.md— check shell snippets for parameter quoting, full flag names, etc.testing.instructions.md— check any code examplesgit-workflow.instructions.md— check command examplesshorthand.instructions.md— check that examples don't themselves use shorthandagent-workflow.instructions.md,readme.instructions.md,releases.instructions.md,repository-specific.instructions.md,update.instructions.md,contributing.instructions.md— check for any code examples and validateWhy a separate issue
The PR #17 audit was already substantial; expanding to other modules would have ballooned the scope. Separate PR keeps blame and review tractable.
Acceptance criteria
instruction-templates/andinstructions/mirror copies updated together (no sync drift introduced)[Unreleased]describing the audit and listing changes per moduleRelated