correct
Turn recurring agent mistakes into structural prevention: architecture, types, actionable lint, then behavior tests.
Skill by matthewyjiang
Install
Works with Claude Code, Claude apps, Codex, Cursor, Gemini CLI, GitHub Copilot, VS Code, OpenCode, Goose, Windsurf. Read the repository before you give any tool access to your data.
npx agentmarket install skill/matthewyjiang-correct
Asks which of your tools to add it to (Claude Code, Codex, Cursor, VS Code, Windsurf, Gemini CLI, Claude Desktop) and shows the change first. See https://agentmart.dev/cli.
Get the weekly diff
The weekly diff: one email a week with new agent tools, notable updates and grade changes, plus security alerts for tools you follow. Double opt-in, one-click unsubscribe.
About
Use cases: productivity, agent-building
Reviews
Write a reviewNo reviews yet. Installed it? Reviews come from verified installers only. How reviews work
Permissions
What this skill can do and what it asks for, from its SKILL.md. Checked Oct 25, 2026.
- Tools
- No tools found
- Secrets
- None asked for
- Install scripts
- None found
Network hosts it talks to 1
codeload.github.com· found in install command
Files it touches
~/.claude/skills/correct
What's inside
- Skill name
correct- Description
- Turn recurring agent mistakes into structural prevention: architecture, types, actionable lint, then behavior tests. Use after a repeated correction or when asked to harden the workflow.
- License
- MIT
- Instructions
- 28 lines in SKILL.md · view on GitHub
Files (1)
- SKILL.md
SKILL.md (first lines)
# Correct Read [the workflow contract](../../references/workflow-contract.md). Derived from `correct` and `principle-encode-lessons-in-structure`. Find recurring mistake classes in relevant commits, reverts, reviews, corrections and repository instructions. Distinguish demonstrated recurrence from a speculative new rule. Prefer the highest-level fix that removes the mistake: 1. Architecture: one state owner, one supported route, unreachable internals; derive shared lists from one source of truth, remove misleading dead examples. 2. Types: make invalid states unrepresentable. Then an actionable lint/check whose error names the supported type, function or file. Ratchet only new violations where the existing baseline cannot reasonably be migrated in scope. 3. Behavior tests: drive the real interface; prove the guard fails on an actual past mistake and passes on the intended behavior. Parent runs local/CI-equivalent checks. 4. Docs last for remaining judgment calls, with a concrete failure example. A new instruction alone is not enforcement. When structure solves it, avoid redundant rules. Implement only fixes within the authorized task. Propose broad policy changes rather than editing AGENTS.md, changing CI gates, granting exceptions or creating commits unasked. If the repo already keeps an enforcement table, update it only when asked or clearly in scope; do not impose a new one globally. Respect repo exception policy. Return each class, evidence of recurrence, prevention level chosen, why stronger prevention was unsuitable, and proof that the check catches the real mistake.
From matthewyjiang/rho-pstack, MIT.
More from matthewyjiang/rho-pstack
- 0.567score
interrogate
matthewyjiang
Optional adversarial cross-model review to challenge a design or find blind spots, using per-call agent models or host delegation.
SkillAFreshness: AClaude Code +9 - 0.567score
how
matthewyjiang
Explain subsystem architecture, runtime flow, placement, ownership, or layering before changing code.
SkillAFreshness: AClaude Code +9 - 0.567score
architect
matthewyjiang
Sketch caller usage, types, signatures and module ownership before nontrivial implementation.
SkillAFreshness: AClaude Code +9 - 0.561score
blast-radius
matthewyjiang
Trace what a change could break beyond its diff and prove the load-bearing safety assumptions through real code.
SkillAFreshness: AClaude Code +9 - 0.536score
principle-boundary-discipline
matthewyjiang
Apply when wiring validation, error handling, or framework adapters.
SkillAFreshness: AClaude Code +9 - 0.536score
principle-build-the-lever
matthewyjiang
Apply to any non-trivial work, not just bulk work: edits, migrations, analyses, checks.
SkillAFreshness: AClaude Code +9 - 0.531score
principle-experience-first
matthewyjiang
Apply when product, UX, or feature-scope tradeoffs come up.
SkillAFreshness: AClaude Code +9
Signals
- Stars
- 0
- Forks
- 0
- Open issues
- 0
- Last commit (this folder)
- Oct 17, 2026 (12 days ago)
From GitHub (fetched Oct 24, 2026).
Details
- License
- MIT
- Repository
- github.com/matthewyjiang/rho-pstack
- First seen
- Oct 24, 2026
- Last checked
- Oct 24, 2026
- Source
- GitHubgithub-graphql
More from matthewyjiang
- 0.567score
interrogate
matthewyjiang
Optional adversarial cross-model review to challenge a design or find blind spots, using per-call agent models or host delegation.
SkillAFreshness: AClaude Code +9 - 0.567score
how
matthewyjiang
Explain subsystem architecture, runtime flow, placement, ownership, or layering before changing code.
SkillAFreshness: AClaude Code +9 - 0.567score
architect
matthewyjiang
Sketch caller usage, types, signatures and module ownership before nontrivial implementation.
SkillAFreshness: AClaude Code +9 - 0.561score
blast-radius
matthewyjiang
Trace what a change could break beyond its diff and prove the load-bearing safety assumptions through real code.
SkillAFreshness: AClaude Code +9 - 0.536score
principle-boundary-discipline
matthewyjiang
Apply when wiring validation, error handling, or framework adapters.
SkillAFreshness: AClaude Code +9 - 0.536score
principle-build-the-lever
matthewyjiang
Apply to any non-trivial work, not just bulk work: edits, migrations, analyses, checks.
SkillAFreshness: AClaude Code +9
Cite this page
correct skill. Agent Mart. https://agentmart.dev/skills/matthewyjiang-correct (last updated 2026-10-24)