Community Specification & Attribution Gate

Submit Specification & Claim Authorship

Contribute verified AI agent behavior guardrails and anti-hallucination code diffs. Approved specifications earn permanent attribution across AgentRules.org, search indexes, and local exported IDE files (.cursorrules, CLAUDE.md, AGENTS.md).

Quality Score45 / 100

12 requirement(s) pending

1. Contributor Attribution & Verification

Display Name *

GitHub Username *

Organization / Affiliation (Optional)

Twitter / X Handle (Optional)

Website / Blog URL (Optional)

I am an official author or core maintainer of this framework/library (qualifies for Core Maintainer verification badge)

2. Specification Metadata & Scope

Quick-Start Framework Invariant Templates

Specification Title *

3. Anti-Pattern & System Directives

Anti-Pattern Headline (What mistake does AI commonly generate?) *

Core System Prompt (Directive for AI Agent) *

Normative Guidelines (RFC-2119: MUST, MUST NOT, ALWAYS, NEVER)

MUST NOT use deprecated legacy APIs or synchronous evaluations.

ALWAYS validate all incoming dynamic parameters before executing database or state mutations.

4. Bad vs Good Code Diff (Anti-Hallucination)

Snippet Title

Language

[BAD] Antipattern Code (AI Common Hallucination)

[GOOD] Production Standard (Verified Compliant)

Validates anti-hallucination diff against AST quality gate

Automated Quality Gate

45 / 100

Blocked by Quality Gate (12)

Author name must be between 2 and 50 characters.
Valid GitHub username is required for attribution verification (e.g. 'octocat').
Rule title must be between 8 and 120 characters.
Rule description must be between 30 and 600 characters explaining context.
Anti-pattern headline must explicitly describe the error/hallucination (15-250 characters).
At least one tag is required (e.g. 'nextjs-15', 'typescript').
System Prompt directive must be at least 30 characters of unambiguous instructions.
Example #1: Bad antipattern code snippet is too short or empty.
Example #1: Good production standard code snippet is too short or empty.
Example #1: Bad and Good code snippets cannot be identical.
Example #1: Must explain why the Bad code fails in production.
Example #1: Must explain why the Good code is the verified standard.

Attribution & Verification Workflow

1Instant Security Gate: Real-time automated audit validates RFC-2119 invariants, blocks prompt injection, and checks code diff integrity.

2Editorial Review: Submissions enter our review queue where maintainers verify production reproducibility and validate author credentials.

3Permanent Author Attribution: Approved rules are published to AgentRules.org and exported to .cursorrules, CLAUDE.md, and AGENTS.md with full attribution.