harden

Harden

MANDATORY: Re-read this entire file before every audit pass. Do not rely on earlier readings within the same conversation. Context window drift causes skipped steps.

Purpose

Ensures substantial deliverables are structurally sound, internally consistent, and aligned with domain best practices before delivery. Replaces ad hoc quality checks with a repeatable methodology.

How It Works

1. Research — Searches the web for domain-specific best practices before auditing (not just internal assumptions)
2. Dual-pass audit — Checks both ideas (are they sound?) and structure (do paths/references/formatting work?)
3. Severity classification — 🔴 Critical / 🟡 Significant / 🟢 Minor with specific locations
4. Fix — Implements all critical and significant fixes (not just reports)
5. Converge — Re-audits until only minor issues remain (max 3 passes)

The Process

Step 1: Research

Search for current best practices specific to the deliverable's domain.

Scope guidance:

• Minimum 2 searches for focused domains (e.g., a skill file -> Anthropic skill docs)
• 3-5 searches for broad or unfamiliar domains (e.g., a grant aims page -> NIH guidelines, successful examples, common reviewer complaints)
• Always check for official/canonical sources first, then practitioner insights
• Fetch the canonical source. If search results include the single most authoritative document for this domain (official guide, spec, standard), fetch and read it — do not rely on search snippets. Snippets capture ~5% of a document's content and miss testing, edge cases, and patterns. Limit to the first ~5000 tokens if the source is very long; prioritize sections on requirements, validation, and common mistakes.
• Synthesize findings into actionable bullets that inform the audit — not a reading list

If web search is unavailable: State this limitation explicitly. Proceed using training knowledge but flag that the audit lacks external validation.