> For the complete documentation index, see [llms.txt](https://docs.cortex.aic.io/llms.txt). Markdown versions of documentation pages are available by appending `.md` to page URLs; this page is available as [Markdown](https://docs.cortex.aic.io/workflow-guides/analyse/analysis.md).

# Analysis

**Open this area:** In the CORTEX menu, select **Intelligence cycle → Analyse**.

Compare hypotheses, retain contrary evidence, and explain probability, confidence, and uncertainty before submitting the exact content for peer review.

**Who uses this:** Analyst, Reviewer, Author\
**Open in CORTEX:** `/analyse`\
**In-app guide:** `/help?topic=analysis`

**Expected result:** A retained analytical revision explains the reasoning and records a separate eligible review decision.

## Before you start

* Have Analyse and Memory enabled, a visible requirement, and current analytical write authority.
* Use handling that covers the requirement and all intended evidence; it remains fixed after creation. Restricted handling also needs a current approved processing activity for the exact purpose and scope.
* Arrange an eligible reviewer who has not contributed to the retained analytical lineage.

## What to do

{% stepper %}
{% step %}

## Frame the assessment

Find workspaces first: the pager appears only beyond one page and shows the authorised range. Editing search or review filters leaves paging on the last applied result until **Find workspaces** is selected.

The Basics step needs a visible requirement, title, question, and context. For unrestricted handling, **Create draft with basics** saves immediately; evidence, hypotheses, and judgements can be added later.

Restricted handling also needs the approved activity in Evidence before the final save.

The coloured step states show what is ready or still needs attention.
{% endstep %}

{% step %}

## Build the evidence board

Resolve exact passages and label their role as context, corroborating, or contradictory.

Record information credibility and assessment notes; keep it separate from the pinned source's reliability.
{% endstep %}

{% step %}

## Compare explanations

Write hypotheses, assumptions, competing explanations, and discriminating checks.

Add judgements with relevant evidence keys, an explicit PHIA probability, separate analytical confidence, and a substantive rationale.

Mark unsupported judgements honestly and retain gaps.
{% endstep %}

{% step %}

## Save and submit

Complete the Review step and select **Save analytical draft**, inspect the retained revision, and resolve unassessed or legacy likelihood choices before selecting **Submit for peer review** with a reason.

At least one hypothesis and judgement are required for submission, though the initial draft can be smaller; under-review content is locked.
{% endstep %}

{% step %}

## Review the exact content

The workspace list names Draft, In peer review, Changes requested, and Peer reviewed states and explains their immediate meaning.

The independent reviewer checks evidence, alternatives, uncertainty, and gaps, then uses **Record peer review** to endorse or request changes.

Peer reviewed applies only to the exact analysis revision; it is not product approval or release.

Amending a reviewed or changes-requested record creates a new Draft revision; arrange fresh review for the amendment.
{% endstep %}
{% endstepper %}

## Check the result

* What competing explanation could also fit the evidence?
* Have source independence and contradictory reporting been considered?
* Is likelihood separate from confidence in the information and reasoning?
* Are unsupported judgements and information gaps explicit?
* Is the reviewer independent of all contributors?

## If you get stuck

<details>

<summary>Why does opening an analytical workspace briefly show Loading?</summary>

CORTEX completes the page shell first, then retrieves the authorised workspace and revalidates its retained evidence after the interactive connection is ready.

Keep the page open while that bounded check runs. If it cannot complete, the page shows an actionable failure and **Retry loading workspace**; do not create a replacement assessment or weaken access checks.

Refresh once after connectivity returns. If the failure persists, give the operator the time, workspace identifier, and displayed reference.

Never copy protected analytical content into an ordinary support channel.

</details>

<details>

<summary>What happens if finding or paging workspaces fails?</summary>

The list keeps the last confirmed results, filters, and page after a temporary failure. The warning identifies them as earlier results.

Select **Find workspaces** to retry edited filters, or select the same page control to retry paging; neither action changes retained analysis.

If access is lost, the protected list and create action are cleared. Restore current authority and refresh before continuing.

</details>

<details>

<summary>Why is submission or endorsement refused for an older likelihood?</summary>

Legacy scale values retain their original meaning.

Deliberately reassess the judgement using a current PHIA term and save a new revision; request changes on an old submission rather than silently relabelling it.

</details>

<details>

<summary>Why can I not review when I have review permission?</summary>

A retained contributor is not independent.

Ask another eligible reviewer with the complete current workspace and evidence scope; administrator status does not bypass the rule.

</details>

<details>

<summary>Why did a reviewed assessment return to Draft?</summary>

The content changed.

The old review remains attached to its exact previous revision; inspect the amendment and submit it again rather than describing the new text as already reviewed.

</details>

## Responsibility and limits

* Analysts explain uncertainty and the basis for confidence; a catalogue term is not a computed probability or quality certificate.
* Reviewers challenge the exact reasoning and its limitations. Review does not satisfy Direction, accept Synapse relationships, or approve product dissemination.

## Reference and scope

JDP 2-00 (4th ed.), 2.18, 3.41–3.48 and 7.7 inform analytical rigour, alternatives, and uncertainty. Requiring assessed confidence at submission is CORTEX policy.

## Related workflows

* [Find evidence and select exact retained passages](broken://pages/e6eab51e146c4db1e002b720557a5faea932e41c)
* [Record cited entities and challenge relationships](broken://pages/0052d222b6bbe9bca3b9b5abb372bd0a965b393a)
* [Author an intelligence Product and inspect its exports](file:///7397340/product/products.md)