correction-response

A rule for handling user corrections by deciding whether they reveal a reusable working principle or only apply to the current task.

In plain words
What is it for?
Use it to acknowledge corrections, ask when their scope is unclear, update rules for recurring issues, and apply relevant changes immediately.
Why use it?
It helps the agent learn from repeated mistakes without turning every one-off preference into a permanent rule.

Cursor rule for Cursor

Install

Getting it into your agent

One page per mod, every tool's command on it. A separate URL per tool would split the same page into five that compete with each other.

agentmods
npx agentmods add rules/langwatch/better-agents/correction-response
Clone the repo
git clone --depth 1 https://github.com/langwatch/better-agents

Made for: Cursor.

Per session 319 This file is loaded in full into every session.
When invoked 319 The same file — it is already loaded in full.
Security scan A 0 findings. Scan, not verified.
Origin original No closer match found in the catalogue.
Token cost

What it costs to keep this loaded

Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.

ModelPer sessionOnce invoked
Fable 5 $0.00319 $0.00319
Opus 5 $0.00160 $0.00160
Sonnet 5 $0.00064 $0.00064
Haiku 4.5 $0.00032 $0.00032

Measured 2d ago against content hash f79efd8130cb, method: parsed. Prices are Anthropic first-party input rates as of 2026-08-30, from the pricing page.

Security

Grade A, and why

correction-response scanned grade A with 0 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 2d ago.

A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.

Nothing flagged

None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.

.cursor/rules/correction-response.mdc · 33 lines

What it actually says

Correction Response Rule

When corrected for doing something wrong, evaluate if it reveals a pattern that should become a rule. Only create/update rules for systemic issues, not one-off instructions.

Process:

  1. Acknowledge the correction and thank the user
  2. Ask for clarification if the correction seems like a one-off instruction vs. a principle
  3. Identify the root cause of the mistake
  4. Create or update a rule only for patterns that should be permanently avoided
  5. Apply the rule immediately to the current work if created
  6. Document the change clearly

Examples of Rule-Worthy Corrections:

  • "no any in typescript" → Create typing rules (systemic principle)
  • "rule too long" → Update rule formatting guidelines (process improvement)
  • "use pnpm not npm" → Update tooling preferences (workflow consistency)

Examples of One-Off Instructions (Don't Make Rules):

  • "skip the test" → Just skip this one test, don't change testing workflow
  • "use red color here" → Specific UI choice, not a principle
  • "name this variable foo" → Specific naming, not a pattern

When to Ask for Clarification:

  • If the correction seems situational or temporary
  • If it contradicts existing patterns/principles
  • If you're unsure if it's a one-off or systemic issue

This ensures meaningful rule creation while respecting situational instructions.

Changes

What this file has done since we first saw it

Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.

  1. 2d ago First seen · 33 lines · 319 tokens per session scan A f79efd8130cb

Subscribe to this mod's changes

correction-response is a cursor rule published in the GitHub repository langwatch/better-agents (1,552 stars, last pushed 3mo ago), licensed MIT. It adds 319 tokens to every session, about $0.0016 per session on Opus 5. A static security scan graded it A with 0 findings. No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-30.