StegVerse SDK for governed AI execution
Project description
STEGVERSE SDK
Execution is not assumed. Execution is admitted.
StegVerse verifies every action before it happens and produces cryptographic proof of that decision.
WHAT IT DOES
You propose an action
→ StegVerse evaluates it
→ Decision: ALLOW | DENY | DEFER
→ If allowed: execution + receipt
Every executed action produces a verifiable receipt.
WHY THIS MATTERS
Traditional flow
AI decides → executes → humans audit later
StegVerse flow
AI proposes → evaluated at commit → executes only if admitted
This eliminates ungoverned execution at the point of irreversibility.
QUICK START
Install
pip install stegverse-sdk
Example
from stegverse import StegVerseSDK
sdk = StegVerseSDK()
result = sdk.submit_intent(
action="deploy.compute",
target="render.cluster",
parameters={"gpu": "A100", "count": 4}
)
print(result["decision"]) # allow | deny | defer
print(result["receipt_id"]) # verifiable receipt
SAFETY STACK
-
Mathematical Gate (GCAT/BCAT)
- Evaluates admissibility at commit-time
- Denies unsafe transitions by default
-
Human Review
- Handles ambiguous edge cases
-
Circuit Breakers
- Stops execution on system instability
-
Consensus Controls
- Multi-party emergency halt
-
Fail-Safe
- Defaults to denial if control is lost
LLM ADAPTER
Govern any LLM output before execution:
from stegverse import StegVerseLLMAdapter, LLMProvider
adapter = StegVerseLLMAdapter()
result = adapter.govern_llm_output(
provider=LLMProvider.OPENAI,
model="gpt-4",
prompt="Write a risk scoring function",
output=llm_output
)
Returns:
- decision (allow | deny | defer)
- receipt
- reasoning
Ecosystem Optimization
result = adapter.optimize_ecosystem(
ecosystem_metrics={"cpu": 0.85, "memory": 0.90},
proposed_changes={"type": "scale", "cost": 5000}
)
THE MODEL
Legitimacy constraint:
Φ(x) = K · g^α · c^β · t^γ − a
Where:
- g = governance capacity
- c = constraints
- a = action pressure
- t = trust
Decision rule:
- Φ(x) ≥ 0 → ALLOW
- Φ(x) < 0 → DENY
INTEGRATION
| Downstream | Consumes |
|---|---|
| TV/TVC | Ephemeral secrets via TrustVault |
| GCAT-BCAT-Engine | Deployment verification |
| demo_ingest_engine | Orchestrated ingestion |
| AaCT-E | Audit trail |
| StegDB | State monitoring |
LINKS
- Docs: https://stegverse.org/docs
- API: https://api.stegverse.org
- Issues: https://github.com/StegVerse-org/StegVerse-SDK/issues
- Ingestion: https://github.com/StegVerse-org/demo_ingest_engine
- Email: sdk@stegverse.org
ONE LINE
StegVerse enforces commit-time governance with verifiable execution receipts.
Project details
Download files
Download the file for your platform. If you're not sure which to choose, learn more about installing packages.
Source Distribution
Built Distribution
Filter files by name, interpreter, ABI, and platform.
If you're not sure about the file name format, learn more about wheel file names.
Copy a direct link to the current filters
File details
Details for the file stegverse_sdk-1.0.13.tar.gz.
File metadata
- Download URL: stegverse_sdk-1.0.13.tar.gz
- Upload date:
- Size: 15.5 kB
- Tags: Source
- Uploaded using Trusted Publishing? No
- Uploaded via: twine/6.2.0 CPython/3.11.15
File hashes
| Algorithm | Hash digest | |
|---|---|---|
| SHA256 |
df57856a351d0f1f00f18f879181ac883d2446f78de4900305ebd39a2c8ed291
|
|
| MD5 |
a6b296b366fc5dfdb42876d56c064e57
|
|
| BLAKE2b-256 |
aaa6a123768a9d3e187069a3119cee20969e958f90151d17a2a2b9e40b1b27f3
|
File details
Details for the file stegverse_sdk-1.0.13-py3-none-any.whl.
File metadata
- Download URL: stegverse_sdk-1.0.13-py3-none-any.whl
- Upload date:
- Size: 13.7 kB
- Tags: Python 3
- Uploaded using Trusted Publishing? No
- Uploaded via: twine/6.2.0 CPython/3.11.15
File hashes
| Algorithm | Hash digest | |
|---|---|---|
| SHA256 |
740cde56544d2cc0b02fd7673bd72ffc40f27c9e323ffaf4f9ddbd456ad45864
|
|
| MD5 |
8af248b143ad331202ca856ec3840197
|
|
| BLAKE2b-256 |
eee67605709c08f9b869eb57fcc5820d15417837ad6c29305097ecd2f60582d3
|