Operator
@Site Audit Operator
The stance, domain, quality bar, and review behaviour that should travel with the job.
- - Critical conversion reviewer
- - Evidence before certainty
- - Prioritised fixes
Loading page
One complete run
Start with the kind of request people already type into chat. MPV keeps it connected to the role, method, context, trace, and reviewed deliverable.
“Audit this landing page and give me the fix list.”
Client workspace
Landing page audit / example run
Output reusable
Thread
Audit this landing page and prioritise the fixes for the next build pass.
Saved output
01Before and after
Normal chat loses
MPV preserves
02Five steps
Landing page audit thread
You
Describe the job and attach the current source material.
MPV
Keeps the request inside the correct project and workspace.
Inspect: Original request and project scope
Ready-to-run context bundle
You
Choose the role, method, and knowledge the task should use.
MPV
Adds the Site Audit Operator, Landing Page Audit playbook, and brand context.
Inspect: Operator, playbook, and three sources
Inspectable run record
You
Confirm the expected output and start the run.
MPV
Checks required context and records the assembled prompt.
Inspect: Readiness and prompt snapshot
Reviewed fix list
You
Inspect sources, events, validation, and unresolved questions.
MPV
Keeps the method visible instead of presenting a magical final answer.
Inspect: Trace, source context, validation, and notes
Landing page audit - prioritised
You
Promote the deliverable only when it is useful enough to reuse.
MPV
Stores the output with provenance and makes it available as future context.
Inspect: Review decision and source run
03Run contracts
One example runs through all three contracts so the relationship stays concrete.
Operator
@Site Audit Operator
The stance, domain, quality bar, and review behaviour that should travel with the job.
Playbook
/Landing Page Audit
The input shape, method, checks, examples, and output contract that make the work repeatable.
Knowledge
#Brand Context
The project-specific source material the run should use instead of asking the model to guess.
04Review decision
The output moves through explicit states before it becomes reusable context.
Generated
Check structure and missing context.
Human checked
Confirm evidence, claims, and usefulness.
Saved
Keep the trace and sources attached.
05Working outputs
These are clearly labelled example runs, not invented customer claims.
Example workspace run
Seven fixes grouped by impact and effort, with the run trace attached.
Source
Screenshot + brand context
Example workspace run
Evidence, trade-offs, recommendation, and open questions in one reviewable document.
Source
Research notes + project decisions
Example workspace run
A reusable angle set, channel adaptations, and the source material behind them.
Source
Offer + ICP + voice guidelines
Run the loop