// the free audit

Find the workflow worth building for free

Paste this into your agent and let it handle the rest. It interviews you to find the highest-leverage workflows, and writes a report explaining what to build and why.

install-prompt.md
Install the AI Workflow Audit for this agent.

The AI Workflow Audit is a free agent that interviews a founder about one messy, repetitive workflow, then produces a build-ready AI Workflow Audit Report: a current-state map, the bottlenecks, an AI opportunity scorecard, a recommended first workflow, a V1 system design, human-review points, risks, and an eval plan. It understands the workflow before it recommends anything.

It is a directory of 13 files, not a single file: SKILL.md plus its templates, rubrics, and examples. SKILL.md links to those by relative path, so install the whole directory or it will run shallow.

If you are Claude Code, install it cleanly with the one command below. It fetches every file into ~/.claude/skills/ai-workflow-audit/ and is safe to re-run:

  curl -fsSL https://raw.githubusercontent.com/ryan-hennebry/ai-operator/main/scripts/install-claude-skill.sh -o /tmp/install-audit.sh && bash /tmp/install-audit.sh remote

If you cannot run shell commands (Codex, Cursor, or a plain chat), do not fake it. Read the agent directly from https://raw.githubusercontent.com/ryan-hennebry/ai-operator/main/skills/ai-workflow-audit/SKILL.md , follow every relative link it references from the same directory, and run the audit from those instructions. If you cannot fetch URLs or write files at all, stop and tell me to switch to Claude Code, Codex, or Cursor.

Before you start, ask me exactly one question:

Which workflow should we audit first?
1. Founder-led GTM and market intelligence (the usual starting point)
2. Customer support or operations
3. Hiring and recruiting ops
4. Something else, I'll describe it

Then begin stage 1. Interview me about my company and that workflow before proposing anything. Do not name a tool, model, or architecture until the workflow is mapped. Finish by producing the full AI Workflow Audit Report and telling me what to build first and why.
  • It runs inside your coding agent and will interview you.
  • It produces a build-ready report you keep, whether we work together or not.
  • It is the only way in. No discovery call, you qualify yourself so we know what's best to build.

If the audit finds something worth building, the build sprint starts from £4,500.

// the report you get

The build-ready report the audit hands you

Fourteen sections, in three groups, from company context to the suggested next step.

Diagnose 01–06
01Company context
02Workflows discussed
03Highest-pain workflow
04Current-state workflow map
05Bottlenecks and failure points
06AI opportunity scorecard
Design 07–12
07Recommended first AI workflow
08V1 system design
09Tools and data required
10Human review points
11Risks and limitations
12Eval criteria
Ship 13–14
13Implementation checklist
14Suggested next step