Cost audit
Finds every call to a model in your repository and prices the patterns that run up the bill.
- Agent loops and polling with no limit
- Paid endpoints with no spending cap
- The same answer paid for twice
npx intentiaiCatch expensive AI code before it ships.
IntentiAI audits every model call in your repository for cost problems, and checks what your coding agent changed before you push.
npx intentiai- Open source, MIT
- Runs on your machine
- No API key
Nutlope/roomGPT1 model call, 5 findings
- Replicate job polled with no deadline and no way to cancel itapp/generate/route.ts:692x per timeout
- No spending cap anywhere on the default pathapp/generate/route.ts:40no ceiling
- Rate limit only exists when two env vars are setapp/generate/route.ts:8no ceiling
- Request body reaches the paid call unvalidatedapp/generate/route.ts:40one call per abuse
- Identical requests regenerate the same imageapp/generate/route.ts:40full price again
Ceiling on this billnone
- 98.6%
- of model calls found across seven hand-labelled open-source apps
- 45%
- of labelled cost problems found in apps the rules were not written against
- 13 of 14
- Ship Check findings correct on 30 recent commits across five languages
Two checks in one command line tool
One for what your AI code will cost. One for what your coding agent just changed.
Ship Check
Reads the change your coding agent just made and flags what a passing test run hides.
- Tests deleted or narrowed until they pass
- Migrations that break the running app
- Webhooks that skip the signature check
npx intentiai checkHow a report reaches you
Run it
Run one command in any repository. You get a summary in your terminal right away.
Open your report
Each finding comes with the file, the line, the fix, and the research behind the rule.
Keep it, or don’t
Sign in to keep reports and track them over time. Unclaimed reports are deleted after seven days.
Your source code stays on your machine
Both commands run locally. The cost audit uploads findings only, never code. Ship Check uploads nothing unless you ask it to.
Read the privacy statement- Cost audit
- Uploads rule, file, line and fix to a private link.
--localuploads nothing. - Ship Check
- Writes an HTML report on your machine.
--shareuploads findings only. - Never sent
- Whole files, diffs, or your git history.
Guides
All guidesHow to reduce LLM API costs: patterns that inflate the bill
The code patterns that inflate an LLM API bill: missing output caps, oversized models, resent history, unbounded loops, no budget guard, and how to fix each.
How to review AI-generated code before you push: a checklist
A practical checklist for reviewing what Claude Code, Codex or Cursor changed before you push: tests, migrations, auth, secrets, dependencies, errors and scope.
How to stop runaway AI agent loops before they burn budget
Why AI agent loops run away, what they cost as context grows, and how to bound them with iteration limits, token budgets, deadlines and repeated-action checks.
Questions
- What does IntentiAI do?
- IntentiAI is an open-source command line tool with two checks. The cost audit (npx intentiai) finds every AI model call in a repository and flags patterns that raise the bill, such as unbounded agent loops or missing spending caps. Ship Check (npx intentiai check) reviews the diff a coding agent produced and flags risky changes like deleted tests or unsafe migrations before you push.
- Does IntentiAI upload my source code?
- No. The cost audit uploads only findings (rule, file path, line, summary and fix) to a private link, and uploads nothing with --local. Ship Check keeps everything on your machine unless you add --share.
- Is IntentiAI free?
- Yes. The CLI is free and MIT licensed, runs locally and needs no API key.
- Which languages are supported?
- The cost audit reads TypeScript, JavaScript, Python, Go, Java, Ruby and PHP. Any other language in the repository is counted and named in the report rather than silently skipped.
Run it on your repository
Free, open source, and nothing to configure.
npx intentiai