ACAAS is in public beta. Behavior may shift between minor versions; breaking changes are called out explicitly. Subscribe to the RSS feed to get notified when anything ships.Documentation Index
Fetch the complete documentation index at: https://acaas.mintlify.app/llms.txt
Use this file to discover all available pages before exploring further.
Rate limit introspection
AddedGET /v1/rate-limits so callers can
inspect their current quota without spending a request. The response
includes requests_remaining, resets_in_seconds, and a categorical
status (ok, approaching_limit, at_limit).Authenticated responses now also include the X-RateLimit-Remaining and
X-RateLimit-Reset headers, so you do not need a separate call to know
where you stand.See the new Rate limiting guide for the full ladder
and proactive pacing patterns.Tunable emphasis ratio
/v1/emphasize now accepts an emphasis_ratio parameter between just
above 0 and 1.0, defaulting to 0.3. Lower values produce sparser
highlights; higher values are more generous.The emphasis model itself was retrained on a corpus of release notes,
product announcements, and headlines. Output is meaningfully more useful
for short text and slightly more verbose for very long input.Behavior remains non-deterministic. See
Experimental emphasis for guidance on when
and when not to depend on this endpoint.Whisper response shape
Fixed a bug where/v1/whisper returned characters_amplified: 0 for
any input that was already lowercase. The field now correctly reports
the count of characters whose case changed, regardless of direction.Fixed a regression where /v1/scream with intensity: 1 occasionally
appended two exclamation marks instead of one. The villain has been
dispatched.Public beta launch
ACAAS is now generally available in public beta. All six endpoints are live:POST /v1/shout— uppercase a string.POST /v1/scream— uppercase with configurable exclamation intensity.POST /v1/whisper— the quiet cousin. Returns lowercase.POST /v1/emphasize— experimental. A model picks which words to bold.GET /v1/health— service health. No auth required.GET /v1/rate-limits— quota introspection.
Authorization: Bearer header to a dedicated X-API-Key header. Update
any pre-launch integrations accordingly.Request bodies are now capped at 10,000 characters. Larger payloads
return 413 Payload Too Large.Start with the Quickstart or jump straight to the
API reference.