Code Review API - AI Pull Request and Diff Reviewer

## AI Code Review API Send a unified diff or raw files, get back structured, machine-readable findings, not a wall of prose. Built for CI pipelines, PR bots, and dev tooling that need to parse review output programmatically. ### What you get - **Structured JSON findings**: `file`, `line`, `severity`, `issue`, `suggestion`, plus a plain-language `summary`. - **Line-verified**: findings are checked against your diff's actual changed lines; anything unverifiable is flagged (`line_unverified`), not silently trusted. - **Tuned reviewer model on every plan** (`crank-reviewer:latest`), fine-tuned specifically for catching real bugs/security issues, not style nits. The free plan is reviewed by the same model as the paid ones. - **Never fails silently**: malformed model output still returns a usable response (`parse_warning` + raw summary) instead of a 500. - **Truncation-safe**: long diffs that hit the token cap are flagged (`truncated: true`) so you know findings may be incomplete. ### Endpoint `POST /v1/review`: pass `diff` (unified diff text) or `files` (raw file contents), get structured finding### Plans Every plan is reviewed by the same tuned model — plans differ on quota and rate limit, not review quality. | Plan | Price | Quota | Rate limit | |---|---|---|---| | BASIC | $0/mo | 100 requests | 10/min | | PRO | $9/mo | 2,500 requests | 20/min | | ULTRA | $29/mo | 15,000 requests | 40/min | | MEGA | $99/mo | 60,000 requests | 60/min | | Monthly PAYG | usage-based | $0.01/request, graduated | 30/min | ghput) | | Mega | crank-reviewer v5 (tuned, B2B/dedicated) | ### Why General-purpose LLM APIs give you prose. This API gives you `file:line` + severity you can wire straight into a PR comment bot, a CI gate, or a dashboard, at a fraction of GPT-4/Claude API cost.