Standard Code Review

What you get

• Branded HTML report— score dashboard, three headline findings, per-category breakdowns with "What works" and "What needs work", full action list (P0/P1/P2/P3 with effort and cost-of-inaction), Decisions Required section, Open Questions section. Print-ready PDF.
• Markdown working copy — same content, diff-friendly, drop-into-Linear-as-tickets.
• findings.json — machine-readable structured output. Imports into Linear / Jira / GitHub Issues. Used by HRC on the next cycle to verify which findings you closed.
• Embeddable score badge — public SVG you can put in your README: ![HRC: B+ (84)](https://...)
• 30-minute walkthrough call — three headline findings, Decisions Required, what to close first.

The 10-category rubric

Weights are fixed across reviews — scores are comparable across customers and across time. Security carries the most weight; functional correctness and reliability are next.

• 1.Functional Correctness15%
• 2.Architecture & Structure10%
• 3.Security18%
• 4.Reliability & Error Handling15%
• 5.Maintainability & Code Quality12%
• 6.Operational Readiness8%
• 7.Portability & Vendor Lock-in7%
• 8.Dependency Hygiene5%
• 9.Performance & Cost5%
• 10.IP & Ownership Clarity5%

Each category has a published 0–100 letter-grade ladder with concrete A-tier / B-tier / C-tier / D-tier signals. Read the full rubric (Field Notes Vol 06) →

How it works

1. 1You buy the review ($2,000 via Stripe). Add your repo URL now or after.
2. 2Daniel emails you within 24 hours to confirm scope and ask for repo access. Install the HRC GitHub App on the specific repo (one click) — or upload a zip if you prefer.
3. 3Within five business days you receive: the HTML report, the markdown, the findings.json, and a calendar invite for the 30-minute walkthrough.
4. 4You ship the fixes. When you're ready for a re-review (any time within 6 months), it's 50% off — $1,000.

Accountability commitment

Every HRC Code Review is anchored to a specific commit SHA. The score and badge represent the codebase at that anchor — not a perpetual warranty. If you ship features after the review, your score still reflects what we actually reviewed; the badge marks itself as stale after 180 days if you don't re-review. We don't vouch for changes we didn't read.

Why HRC and not a free LLM tool?

OpenAI, Anthropic, and Google could each ship a free code-review tool tomorrow. Some eventually will. Here's what their version structurally can't be — three short answers; the long-form essay version is “The Sycophancy Tax” in Field Notes Vol 07.

• 01
  A real engineer signs your report. Daniel personally finalizes the deliverable, runs the walkthrough call, and stands behind it. Hyperscalers ship products at scale; they don't run consulting ladders.
• 02
  The methodology is operating procedure, not just a rubric. 50+ pages of internal adjudication procedure — tone rules, score-cap anchors, decisions-vs-findings split — refined across every review we ship.
• 03
  Our calibration data compounds. Every refund on the $300 tier teaches the system. After 100 reviews, the labeled false-positive dataset becomes proprietary calibration data no hyperscaler has — and we ship paid reviews under that improving calibration.