One-time diagnostic

AI Stack Audit

CA$1,000

A higher-value cross-layer review for an AI-generated prototype or small repo: generated code, configuration, dependencies, tooling, deployment path, and readiness seams.

After delivery, clients may provide private feedback and optionally approve a public testimonial.

What this probe helps you decide

Whether an AI-generated prototype or small repo has real enough support layers to demo, hand off, repair, pause, or scope for deeper stabilization.

Best lifecycle moment

When an AI-generated prototype or small repo needs support-layer review across the stack.

What you submit

One AI-generated prototype or small repo, the product/surface it is meant to support, its intended use, known constraints, setup/deployment notes if safe to share, and the decision the audit should support. Do not submit secrets, credentials, or private source code through intake.

What product or surface this targets

Tell us the product, service, feature, SDK, app, workflow, dependency choice, AI-generated slice, or legacy module inside the repo/project that the probe should focus on. The repo is evidence; it is not automatically the whole target.

What we inspect

AI Code Stability signals plus dependency reality, generated architecture, test depth, configuration/runtime assumptions, tooling, deployment path, observability posture, and top false-closure risks. Evidence is bounded to the accepted surface and safe submitted context.

What the report returns

CA$1,000 buys a broader review of an AI-generated prototype or small repo. The report includes:

  1. Verdict — promote, hold, sandbox, repair, or seek quote
  2. Stack evidence map — generated code, dependencies, config, tooling, deploy/runtime, tests, and docs
  3. Generated-code seam findings — false-closure seams that make the stack look more complete than it is
  4. Dependency/tooling risk notes — unsupported packages, missing lockfiles, stale toolchains, or unclear setup paths
  5. Configuration/runtime gaps — missing environment, deployment, observability, auth, or failure-mode evidence
  6. Boundary of claim — what was and was not checked
  7. Next smallest stabilizing move — the first action likely to improve the stack
  8. Optional Stability Ledger entry — how this audit changes the private continuity record

How this differs from related probes

  • Free First Probe routes the request; AI Stack Audit delivers the cross-layer report.
  • AI Code Stability checks a focused generated-code slice for false closure. AI Stack Audit widens to config, dependencies, tooling, runtime, deployment, and support-layer readiness seams.
  • Paid Trial Diagnostic answers a broader next-step project question. AI Stack Audit is specifically for AI-generated prototypes or small repos where the whole support layer needs review.
  • Stability-Signal checks stale trust and support reality. AI Stack Audit checks whether a generated stack can be safely stabilized or handed off.
  • Dependency Watch is recurring. AI Stack Audit is a one-time deeper cross-layer review unless follow-up cycles are separately scoped.

Stability Ledger

An AI Stack Audit can append a private Stability Ledger entry for the scoped prototype or small repo. The entry records the stack evidence map, supported claims, missing evidence, unresolved seams, next stabilizing move, and whether the recommended next step is repair, sandbox, handoff, another probe, or Custom / Quote. It is a client-facing continuity artifact, not a public registry, certification, guarantee, or internal tracker export.

Sample report / example

See the AWS Pydantic Agents AI Stack Audit sample: a support-layer review that turns honest boundary language into a readiness checklist.

What this does not include

  • Implementation
  • Security audit
  • Production certification
  • Unlimited advisory support

Boundary of claim

This audit reports visible support and gaps across a small scoped surface. It does not fix the stack or certify it for production.

After delivery

Reports are delivered as owner-review-pending. After review you may act on the stabilization backlog, ask for clarification, request a narrower AI Code Stability follow-up, or scope Custom / Quote work. FoldEngine does not publish findings without explicit client approval.

Best-fit examples

  • AI-generated prototype before demo or handoff.
  • Small repo with generated architecture and unclear wiring.
  • Prototype that needs a stabilization backlog.