Nothing to hand? Load the
— a literal API key, a model-supplied string concatenated into a SQL query,
MemorySaver, an undocumented tool, an ungated send_email and a
while True: loop around .invoke() — or the
,
where the correct verdict is ship-ready and the useful output is what to add next.
Paste the agent — the prescan is free
No upload, no AI: the prescan reads your source in the browser and lists what it mechanically found. The inventory of agents, sub-agents, tools, prompts, middleware, checkpointers, stores and graphs, then the flags — secret-looking literals, tools with no docstring or description, destructive tools with no approval gate, thread_id used with no checkpointer, in-process persistence (MemorySaver, InMemoryStore), raw host shell access, exceptions swallowed inside tools, unbounded agent loops, user input interpolated into the system prompt, fixed sleeps in agent or tool code, tools returning an unbounded payload, and model snapshots pinned inline. Each group explains why it matters. This part costs nothing and happens while you type.
The AI reviews it — this is the metered part
A senior agent-infrastructure engineer's pass: a production-readiness posture with the single most important change named, the inventory with each construct's role, and prioritized findings across correctness, context, tools, safety, reliability and hygiene — each with the problem, the concrete impact in production and on the team, the fix and a corrected Python or TypeScript fragment in your stack's language. Every prescan flag is confirmed or explicitly set aside. Findings may only cite agents, tools, sub-agents, prompts and files that actually appear in your code. Pricing is honest: a worst-case amount is reserved before the run and only what the run actually uses is charged — the meter next to the button shows both.
Fix, export, re-run
Every corrected snippet in one paste-ready block, the findings as ready-to-paste GitHub PR review comments, each one headed with the construct it is about and carrying its fix in a ```suggestion fence — paste a block on that construct's line in the diff and GitHub offers it as a one-click commit; a tickable action checklist for the ticket, the findings table as CSV, and Markdown or JSON export of the whole review. Then start fixing: the prescan re-runs as you edit and the strip above the run button counts the flags you have cleared, the ones still open and any you have just introduced — in the browser, for free, before you pay for a second review. Reviews are saved to your SkillSafe account when you are signed in, so they follow you to another machine; restore puts the code back in the form too.
Derived from the @langchain-ai/deepagents skill (MIT license).