Settings & Plan

The settings endpoints control how Aurra behaves for your API key (Level 2 auto-supersession, excluded categories, confidence floor, classifier model). The plan endpoint reports usage against your quota.

GET `/agent/settings`

Read the current settings for your API key.

Example

curl https://api.aurra.us/agent/settings \
  -H "Authorization: Bearer $AURRA_API_KEY"

Response (200)

{
  "auto_supersede_default": false,
  "excluded_categories": ["health_medical", "legal_status"],
  "min_confidence": 0.85,
  "classifier_model": "claude-haiku-4-5-20251001"
}

Response fields

Field	Type	Default	Notes
`auto_supersede_default`	boolean	false	If true, every write runs the Level 2 auto-supersession classifier. Can be overridden per-request via `auto_supersede` on `/agent/memories`.
`excluded_categories`	array	`["health_medical", "legal_status"]`	Categories that never reach the LLM classifier. Hard-coded deterministic gating.
`min_confidence`	float (0-1)	0.85	Confidence floor. Classifier decisions below this are rejected.
`classifier_model`	string	`claude-haiku-4-5-20251001`	Model used for supersession classification. Must be an Anthropic model ID.

PATCH `/agent/settings`

Update one or more settings. Omit fields you don't want to change.

These are per-API-key defaults. They apply to every write made with this key unless overridden per-request. If you're integrating Aurra into a product, consider issuing separate API keys per environment (dev/staging/prod) so you can test auto-supersession in dev before enabling it in prod.

Request body

All fields optional. Only fields present in the body are updated.

Field	Type	Notes
`auto_supersede_default`	boolean	Toggle Level 2 on/off for this key.
`excluded_categories`	array	Replaces the entire list. Pass `[]` to allow all categories (not recommended).
`min_confidence`	float (0-1)	Raise for stricter gating, lower for more supersessions.
`classifier_model`	string	Must be an Anthropic model ID. Changing this impacts cost and latency.

Example: enable Level 2 auto-supersession

curl -X PATCH https://api.aurra.us/agent/settings \
  -H "Authorization: Bearer $AURRA_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{"auto_supersede_default": true}'

Example: raise confidence floor

curl -X PATCH https://api.aurra.us/agent/settings \
  -H "Authorization: Bearer $AURRA_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{"min_confidence": 0.90}'

Response (200)

Returns the full settings object after update (same shape as GET).

GET `/plan/status`

Current plan tier, usage, and block status. Check this before long-running write-heavy jobs.

Example

curl https://api.aurra.us/plan/status \
  -H "Authorization: Bearer $AURRA_API_KEY"

Response (200)

{
  "org_id": "2a8bfe51-e7ed-4f6e-8cfd-c7062e2a47bf",
  "plan_tier": "starter",
  "plan_status": "active",
  "writes_used": 49,
  "writes_limit": 50000,
  "writes_remaining": 49951,
  "period_start": "2026-05-01T00:00:00+00:00",
  "in_grace_period": false,
  "blocked": false
}

Response fields

Field	Type	Notes
`org_id`	UUID	Your org identifier.
`plan_tier`	string	`starter` (`$29/mo`) or `pro` (`$99/mo`).
`plan_status`	string	`active`, `past_due`, `canceled`.
`writes_used`	integer	Writes consumed this period.
`writes_limit`	integer	Period quota.
`writes_remaining`	integer	`writes_limit - writes_used`. Computed server-side.
`period_start`	ISO 8601	Start of current billing period. Usage resets to 0 at every new period.
`in_grace_period`	boolean	`true` if you've exceeded quota but are still receiving reads. Writes are 429.
`blocked`	boolean	`true` if reads are also blocked (end of grace period without upgrade).

Using plan status

Check plan/status before bulk imports: if writes_remaining is less than your planned batch, the import will partially fail with 429s. For predictable alerting, poll every few hours and notify your team when writes_remaining / writes_limit drops below 0.1.

On this page