Endpoint Specification
Request Headers
string
required
Your secret Curtly API key (
Bearer ctly_live_...).string
Overrides API key AI compression setting for neutral mode keys.
true: Activates AI Inception Mercury 2.5 diffusion for deep semantic condensation.false: Runs deterministic 0ms Safe Vault algorithmic pruning only.
string
Specifies the target compression ladder level:
fast: 0ms overhead, deterministic cleaning and code license stripping.balanced: Default, balanced 40% to 60% token reduction.aggressive: Deepest token compaction (65% to 80% reduction).
Request Body
string
required
The raw text, markdown content, or prompt string to be compressed.
string
default:"balanced"
Compression level:
conservative, balanced, or aggressive.string
default:"openai-o200k"
Target tokenizer model for accurate BPE token telemetry:
openai-o200k(GPT-6 Astra, GPT-5.6, GPT-4o)claude-bpe(Claude Opus 5.5, Claude Fable 5.1, Claude 4 Sonnet)deepseek-llama(DeepSeek-V4.1-Flash, DeepSeek R1, MiMo-V2.6-Pro, Llama 4)universal-gemini(Gemini 3.8 Flash, Gemini 3.8 Live, Grok 4.7, Qwen 3.8)openai-cl100k(Legacy GPT-4, GPT-3.5)
boolean
default:"true"
Isolates and syntax-locks fenced and inline code blocks inside the Safe Vault.
boolean
default:"true"
Preserves JSON objects, schemas, and API payloads while stripping
nulls and empty arrays.boolean
default:"true"
Appends Curtly’s YAGNI Brevity Governor to eliminate up to 80% of downstream LLM generation fluff.
Response
boolean
Indicates whether the compression pipeline succeeded.
string
The final compressed prompt text with all protected placeholders rehydrated and Brevity Governor attached.
object