MIRENA · Product documentation
Give MIRENA the minimum authorised evidence needed for the job.
The useful input pack is not the largest pack. It is the set of current, relevant, permitted, and clearly scoped material that can support the selected workflow and output.
Verified facts and boundaries.
These fields identify the entity, product, document, or operational boundary without turning related parties into the same entity.
Source context comes first.
Source context identifies the site, offer, audience, region, allowed and blocked topics, existing pages, protected assets, brand rules, claims, and desired outcome.
It keeps a page or cluster inside the site's real business and entity universe. A keyword list or competitor export does not replace that boundary.
Input requirements change by workflow.
| Workflow | Minimum required | Useful additions |
|---|---|---|
| Topical mapping | Source context, topic or offer, audience, site goal | Sitemap, URL inventory, queries, competitors, priorities |
| Content brief | Approved page role, user job, audience, source context | Map row, proof, SERP examples, links, brand rules |
| Drafting | Approved brief and source material | Examples, product records, legal language, approved links |
| Rewriting | Current page or draft, repair goal, protected content | Performance evidence, proof, route map, known failures |
| Internal linking | Source and destination inventory, page roles | Crawl export, relationship rules, user journeys |
| Review | Exact asset version, acceptance criteria, evidence | Rendered screenshots, device checks, implementation notes |
Evidence needs identity and limits.
A source should be attached to the claim or decision it supports. Include the publisher, date, relevant passage or field, market or version, and known limitation.
A current source can still be wrong for the page if it describes another product version, plan, country, audience, or time period.
Use only material you are permitted to provide.
- Do not bypass access controls or submit unlawfully obtained content
- Do not copy or upload protected third-party material beyond a lawful and necessary analytical use
- Confirm client, employer, partner, and confidentiality permissions
- Remove credentials, secret keys, passwords, and authentication tokens
- Avoid full payment-card data and unnecessary personal data
- Do not send regulated or highly sensitive data unless the service, platform, contract, and purpose are suitable
ChatGPT data controls remain separate.
MIRENA runs in ChatGPT. OpenAI states that GPT builders cannot view individual conversations. Whether conversations can be used for model improvement depends on the user's plan and data-control settings.
Temporary Chats have separate retention and training behaviour under current OpenAI guidance. If a GPT uses an action or external API, relevant input may be sent to that provider under a separate privacy notice.
Quality states for an input pack.
| State | Meaning | Action |
|---|---|---|
| Ready | Required source, scope, rights, date, and goal are present | Run the selected workflow |
| Qualified | Useful but incomplete or uncertain material | Run only the bounded analysis and preserve uncertainty |
| Needs evidence | A required source or owner is missing | Request the missing record |
| Blocked | Rights, sensitive data, identity, or safety cannot support processing | Do not proceed |
| Stale | Material may no longer describe the current product, page, market, or policy | Reverify before use |
The input pack is part of the audit trail.
The accepted output should identify which inputs were used, which were rejected, which remained unresolved, and which later reviewer or system must confirm the result.
Check the controlling first-party and official sources.
These routes support the current public statement. External platform and legal sources remain subject to their own updates.
MIRENA inputs
Current first-party input requirements and safety notes.
OpenAI data controls
Official current ChatGPT training and Temporary Chat controls.
GPTs in ChatGPT
Official GPT conversation and external API boundary.
Move to the page that owns the next question.
Each route has a separate job so company, product, trust, legal, and compliance information does not collapse into one promotional page.
Owned route
How MIRENA works
Route the input to the correct workflow.
Open route →
Owned route
Outputs
Match the input pack to the expected deliverable.
Open route →
Owned route
Workflow
Review stages, gates, and handoffs.
Open route →
Owned route
Data privacy overview
Website, ChatGPT, provider, and user-control boundaries.
Open route →
Owned route
Acceptable Use
Permitted inputs and prohibited uses.
Open route →
Owned route
Open MIRENA
Start with an approved input pack.
Open route →