> ## Documentation Index
> Fetch the complete documentation index at: https://www.composer.id/llms.txt
> Use this file to discover all available pages before exploring further.

# Score and manuscript

> A destination receives the score, a valid request with the Intent ID. The manuscript behind the decision is served separately to those entitled to it.

Every workforce decision produces two things, and composerID keeps them apart on purpose. This page defines the score and the manuscript, the path each takes, and the one identifier that joins them. Read it if you are reviewing what composerID sends where.

* The **score** is what a destination system receives: a valid request carrying the Intent ID. It is enough for the destination to do its job and nothing more.
* The **manuscript** is everything behind that request: the scored diagnostic, the answers, the ranked outcomes, the documents, the workflow and its signed PDF. It belongs to the programme and is served by the Manuscript API to the programme's managed service provider (MSP) and its reporting suites.

```mermaid theme={"system"}
flowchart LR
  T["Triage decision"] -->|"Intent ID minted"| C["composerID"]
  C -->|"Path 1: score (Intent ID + valid request)"| D["Destination systems"]
  T -->|"Path 2: manuscript (assessment, documents, workflows, signed PDF)"| M["Manuscript API"]
  M -->|"per-customer API keys"| R["MSP and reporting suites"]
  D -.-|"one Intent ID, both sides"| M
```

## Path 1: destinations get the score

When composerID publishes into Beeline, SAP Fieldglass, Coupa or any other destination, the record carries the Intent ID and the fields the destination needs to be valid. It never carries diagnostic content.

<Check>
  The boundary is structural rather than procedural. The payload an adapter publishes from contains only request fields, so there is no configuration path from an answer or a score to a destination field.
</Check>

Every destination page in this documentation states what is sent, what is read back and what is never sent, built from the destination's own create schema. Start from the [destinations index](/destinations/index).

## Path 2: the MSP gets the manuscript

The programme's MSP needs the evidence: which answers led to the decision, how the options scored, what was overridden and why. That evidence lives in Deployed's production backend (WorkAuthor) and is served by the [Manuscript API](/api-reference/manuscript/overview) at `https://deployed.workauthor.com/api/v1`: assessments, documents, surveys and workflows, with signed webhooks for change. Access is granted per programme with per-customer API keys.

|                          | Score                                              | Manuscript                                                                    |
| ------------------------ | -------------------------------------------------- | ----------------------------------------------------------------------------- |
| Who receives it          | Destination systems                                | The MSP and its reporting suites                                              |
| What it holds            | A valid request plus the Intent ID                 | Scored assessment, answers, ranked outcomes, documents, workflows, signed PDF |
| Served by                | The Intent API, through the destination adapters   | The Manuscript API                                                            |
| Carries the Intent ID in | The destination's client-supplied identifier field | `externalId` on assessments and documents (64 characters)                     |

## One identifier joins the two

The Intent ID is written into the manuscript's `externalId` when the decision is made, and into the destination record when it is published. Ask either side with the one identifier and you reach the other half: the score in the system of record, the manuscript behind it, and one reference between them, with no lookup table.

<Warning>
  The two paths are an invariant, not a preference. No diagnostic payload ever crosses into a destination system, and the Manuscript API is never exposed to a destination.
</Warning>

## Next steps

<Columns cols={3}>
  <Card title="The Compliance File" icon="shield-check" href="/concepts/compliance-file">
    The evidence pack assembled from both paths under one Intent ID.
  </Card>

  <Card title="Manuscript API" icon="book-open" href="/api-reference/manuscript/overview">
    Assessments, documents, surveys, workflows and signed webhooks.
  </Card>

  <Card title="Destinations" icon="layers" href="/destinations/index">
    What each destination is sent, reads back and never receives.
  </Card>
</Columns>