# Request enrichment

**Published under `FLEX-DEC-2026-012`.** What the evaluator adds to a request
before policy sees it, which value wins where the two disagree, and which digest
a consumer can reproduce.

Nothing here was previously written down. Both defects the decision record covers
were consequences of that: a consumer could not compute the digest it was told to
compute, and a caller could contradict a registry fact and be believed.

## The evaluator overlays registry facts

`POST /v1/check` receives a `CheckRequest`. Before policy evaluates it, the
evaluator resolves the subject and the resource in the registry snapshot and
overlays what it finds:

| Target | Keys the registry contributes |
| --- | --- |
| `subject` | `type`, `tenant`, and attributes `display_name`, `organization_relation`, `roles`, `groups`, plus every subject claim and metadata key |
| `resource` | `type`, and attributes `path`, `parent`, `labels`, `trust_zone`, `owner`, plus every manifest attribute |

Separately, and before either lookup, a non-empty top-level `tenant` propagates
into `subject.tenant` and `resource.tenant` when those are empty.

An unregistered subject or resource is enriched with nothing. `diagnostics`
reports `registry_subject` and `registry_resource` for exactly this reason.

## The registry wins

**A registry fact is an authority statement. The same key on the request is the
caller's proposal about itself.** Where they disagree, the registry's value
reaches policy and the caller's is discarded.

This was the other way round until 2026-09-07, and it made every registry ceiling
and allowlist advisory. Three verified escalations against the shipped
`ops-warden` package, each a single added key on an otherwise-denied request:

| Sent by the caller | Registry says | Was | Now |
| --- | --- | --- | --- |
| `resource.attributes.max_ttl_hours: 99` | `8` | `allow` a 12h certificate | `deny ttl_out_of_bounds` |
| `resource.attributes.allowed_principals: [root, …]` | `[platform, root]` | `allow` | `deny disallowed_principal` |
| `resource.attributes.allowed_subjects: [<caller>]` | `[platform-steward, …]` | `allow` | `deny unknown_subject` |

The third is the one to read twice: a subject the registry does not know
authorized itself by naming itself in the allowlist it was being checked against.

`internal/decision/enrichment_precedence_test.go` asserts the registry value
reaches policy on a request that contradicts it, against attribute keys real
packages branch on — a test over an unused key would pass while the escalation
stayed open.

### Overrides are reported, not silent

`diagnostics.registry_overrode` lists the keys whose caller-supplied value was
displaced, as `subject.<key>` or `resource.<key>`. Empty on an honest request.

A registry that quietly discards a contradicting claim hides that a caller
asserted authority it did not have. The decision is correct either way; the
attempt is worth seeing.

## The residual: keys the registry does not define

**A caller-supplied attribute for a key the registry does not define still
reaches policy.** This is stated rather than fixed.

So a package branching on an attribute its manifest omits is reading caller
input, and a manifest that stops declaring a key silently hands that key to the
caller. Registering the resource is not sufficient — the specific key must be
present in the manifest.

The structural fix is that a policy should be able to tell a registry fact from a
caller assertion at all, which one merged `attributes` map makes impossible.
`FLEX-WP-0025` carries it. Until then, **a policy that enforces a ceiling or an
allowlist MUST read a key its manifest declares**, and that is a review
obligation on every package.

## Package attribute-read audit (FLEX-WP-0025-T02)

Read every `input.{resource,subject}.attributes.<key>` (including
`object.get`) in the published packages. A ceiling or allowlist must be
declared on every resource/subject that package can be asked about.
First-class registry fields that enrichment copies into attributes (`roles`,
`groups`, `labels`, `trust_zone`, `owner`, `path`, `parent`, plus subject
`metadata`/`claims`) count as declared.

Classification: **ceiling/allowlist** must be facts; **caller-proposed** is
genuinely the request; **ambiguous** is the T01 input.

| Package | Key | Kind | Role | Declared |
| --- | --- | --- | --- | --- |
| ops-warden | resource.max_ttl_hours | ceiling | ceiling | yes (manifest + snapshots) |
| ops-warden | resource.allowed_subjects | allowlist | allowlist | yes |
| ops-warden | resource.allowed_principals | allowlist | allowlist | yes |
| ops-warden | resource.actor_id | identity | allowlist | yes |
| ops-warden | resource.actor_type | identity | allowlist | yes |
| ops-warden | resource.security_zone | posture | ceiling | yes (snapshots; added to `resource_manifest.yaml` 2026-09-14) |
| ops-warden | resource.security_zone_admission | posture | ceiling | yes |
| ops-warden | subject.actor_type | identity | allowlist | yes (`metadata.actor_type`, copied into attributes) |
| railiance-platform | resource.max_ttl_seconds | ceiling | ceiling | yes |
| railiance-platform | resource.allowed_actor_types | allowlist | allowlist | yes |
| railiance-platform | resource.allowed_purposes | allowlist | allowlist | yes |
| railiance-platform | resource.allowed_delivery_modes | allowlist | allowlist | yes |
| railiance-platform | resource.grant_id | identity | allowlist | yes |
| secrets-engine | — | — | no attribute reads | — |
| tenant-engine | — | — | no attribute reads | — |
| qonto-assistant | — | — | no attribute reads | — |
| user-engine | subject.roles | allowlist | allowlist | yes (first-class `roles`; snapshot now names a registration-applicant) |
| user-engine | subject.issuer | allowlist | allowlist | yes (`claims.issuer` on that subject) |
| markitect (example) | resource.labels | classification | allowlist | yes (first-class `labels`) |
| markitect (example) | subject.groups | membership | allowlist | yes (first-class `groups`) |
| markitect (example) | subject.roles | membership | allowlist | yes (first-class `roles`) |
| informed-decision-t03 (fixture) | subject.assurance, groups, principal_type_source, tenant_source | mixed | see T01 | fixture-only; not a published consumer package |

No published-package ceiling or allowlist remains unbacked after the two
declaration fixes above. `ttl_hours`, `purpose`, `requested_ttl_seconds`, and
`principals` are **context** fields (caller-proposed) and are not in this
table.

## Chosen input shape (FLEX-WP-0025-T01)

**Keep the merged `attributes` map. Add parallel namespaces. Do not break the
six packages.**

```text
input.resource.attributes     # merged view (status quo)
input.resource.registry.*     # registry facts only
input.resource.asserted.*     # caller-supplied only
input.subject.attributes      # merged view (status quo)
input.subject.registry.*      # registry facts only
input.subject.asserted.*      # caller-supplied only
```

Why not stop merging: three of the six published packages (`secrets-engine`,
`tenant-engine`, `qonto-assistant`) read no attributes at all; the other three
already have their ceilings declared. A breaking change across six packages
buys nothing on day one and invites a silent miss on a package that still
reads `attributes` by habit. The merged view stays so existing rules keep
their meaning; a ceiling rule **opts in** to `registry.*`.

**Vocabulary collision on `subject.type`:** do not overwrite it with the
CARING registry type (`Human`/`Agent`/`Automation`/`Service`). Packages
compare against the protected system's actor vocabulary (`service`, `adm`,
`agt`, `atm`). Put the CARING type at `input.subject.registry.type` (and the
caller value at `input.subject.asserted.type`). `input.subject.type` remains
the request's actor vocabulary, filled from the caller, defaulting only when
empty — the exception already in `enrichSubjectRef`.

Ambiguous set that T03's validate check must surface, because intent cannot
be inferred from the key name:

- `subject.roles` / `subject.groups` — first-class registry facts when the
  subject is registered; caller input when it is not
- `subject.issuer` — allowlist for public registration; a fact only if the
  directory wrote it
- `subject.type` — two vocabularies, one field (handled above, not by
  precedence)

Package changes wait until T03 makes "read a declared registry key" a
`flex-auth validate` finding rather than a review note.

## Which digest a consumer can reproduce

| Field | Over | Consumer-computable |
| --- | --- | --- |
| `binding.request_digest` | the **enriched** request | **No** |
| `binding.submitted_request_digest` | the request **exactly as sent** | **Yes** — this is the §6.4.2 replay test |
| `binding.approval_binding_digest` | the enriched request minus `context.approval` | No, and it does not need to be |

`request_digest` was published as the replay test for consumers. It is not one:
the evaluator hashes the enriched request, the registry is flex-auth's, and a
consumer recomputing over what it sent gets a different value on every request
whose subject or resource the registry knows. `secrets-engine` found this on its
first live request.

**Use `submitted_request_digest` for §6.4 obligation 2.** Nothing is lost.
Enrichment is a function of the request and the registry snapshot, and
`provenance.registry_snapshot_digest` already pins the snapshot — so the
submitted digest together with that digest identifies the evaluated request
completely. Registry staleness between issue and replay is bounded by the
decision's own lifetime and by §9.7.2's per-input-class visibility deadline,
which is the mechanism that already covers it.

`request_digest` keeps its value and meaning: it pins what the decision was
actually a function of, and it is what makes a decision replayable **by
flex-auth** for audit. It is not a consumer check and is now documented as not
being one.

`approval_binding_digest` needs no submitted form. It is compared to
`claim.binding.pdp_digest`, which `approval-engine` recorded from flex-auth's own
output at issue time — a flex-auth value against a flex-auth value, consistent on
both sides. **A consumer must not compute either side of that comparison
itself**; doing so reintroduces exactly this defect on the dual-control path.