DEVELOPER ALPHA
ONE ENDPOINT.
A NEW DATA BOUNDARY.
Run the gateway locally, keep its Vault in your trust domain, and configure an upstream only after inspecting the protected context.
01 / RUN
docker compose up --build02 / POINT
OPENAI_BASE_URL=http://127.0.0.1:8787/v103 / VERIFY
curl http://127.0.0.1:8787/healthzOPENAI-COMPATIBLE
Existing clients.
Protected messages.
from openai import OpenAI
client = OpenAI(
base_url="http://127.0.0.1:8787/v1",
api_key="local-dev"
)
response = client.chat.completions.create(
model="configured-upstream",
messages=[{"role": "user", "content": private_context}],
extra_body={
"shreddar_task": "Review termination obligations",
"shreddar_policy": "strict-v1"
}
)
print(response.model_extra["shreddar"]["receipt"])METHODENDPOINTJOB
POST
/v1/analyzeClassify risk and preview policy actions.POST
/v1/protectCompile raw task context inside your boundary.POST
/v1/reconstructResolve authorized current-session placeholders.POST
/v1/chat/completionsProtect, infer, inspect and STITCH.POST
/v1/shardCreate semantic compartments for local routing.GET
/v1/receiptsInspect privacy-safe execution evidence.The public playground is browser-local. The gateway running behind https://app.shreddar.pro is authenticated and reserved for operator verification; it is not anonymous hosted private-data processing. Non-streaming chat completions only. Multi-tenant RBAC, PDF/DOCX extraction, live x402 settlement, and onchain anchoring are not represented as complete.
Review security and limitations →