simpleaudit-studio 0.2.2__tar.gz → 0.2.3__tar.gz

{simpleaudit_studio-0.2.2 → simpleaudit_studio-0.2.3}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.5
 Name: simpleaudit-studio
-Version: 0.2.2
+Version: 0.2.3
 Summary: Production AI model audit platform — one-liner demo mode via uvx
 License: MIT
 License-File: LICENSE
@@ -52,7 +52,7 @@ A self-hostable platform for running reproducible AI model audits using the [Sim
 uvx simpleaudit-studio
 ```
 
-[`uvx`](https://docs.astral.sh/uv/#uvx) installs the [`simpleaudit-studio`](https://pypi.org/project/simpleaudit-studio/) package and runs it in an isolated environment. No Docker, no Postgres, no manual setup. Opens at http://localhost:8000 (login: `admin` / `admin12345`). A mock model server is pre-seeded so you can start exploring audit results immediately.
+[`uvx`](https://docs.astral.sh/uv/#uvx) installs the [`simpleaudit-studio`](https://pypi.org/project/simpleaudit-studio/) package and runs it in an isolated environment. No Docker, no Postgres, no manual setup. Opens at http://localhost:8000 (login: `studio` / `admin123`). A mock model server is pre-seeded so you can start exploring audit results immediately.
 
 🌐 Or skip the setup entirely — try the live demo: <a href="https://sushantgautam-simpleaudit-studio.hf.space" target="_blank" rel="noopener noreferrer">sushantgautam-simpleaudit-studio.hf.space</a>
 
@@ -63,7 +63,7 @@ The pre-seeded connections point at a built-in mock server. To run real audits,
 | Provider | Base URL | Example |
 |----------|----------|---------|
 | Ollama (local) | `http://localhost:11434/v1` | Qwen 3, Llama 3.3, Mistral |
-| vLLM (GPU) | `http://your-gpu-server:8000/v1` | Any HF model |
+| vLLM | `http://your-vllm-server:8000/v1` | Any HF model |
 | OpenAI | `https://api.openai.com/v1` | GPT-4o, o3 |
 | Together AI | `https://api.together.xyz/v1` | Many open models |
 | Groq | `https://api.groq.com/openai/v1` | Fast inference |
@@ -72,7 +72,7 @@ Set the API key in the connection form (leave blank for local servers that don't
 
 ## 🐳 Self-Hosting
 
-For teams, multi-user setups, or GPU worker pools, use Docker Compose:
+For teams or multi-user setups, use Docker Compose:
 
 ```bash
 git clone https://github.com/SushantGautam/SimpleAuditStudio
@@ -84,7 +84,7 @@ docker compose up -d
 
 Services: Web UI (:8000), PostgreSQL, Hatchet queue (:8888), Worker. Optional profiles: `--profile storage` (MinIO), `--profile mock` (mock model API).
 
-See [docs/deployment.md](docs/deployment.md) for production hardening, GPU workers, backups, and upgrades.
+See [docs/deployment.md](docs/deployment.md) for production hardening, backups, and upgrades.
 
 ## ✨ What You Can Do
 
@@ -111,7 +111,7 @@ Browser → Django/Gunicorn (:7860)
 
 - [Architecture](docs/architecture.md) — system design, service boundaries, deployment topology
 - [Domain Model](docs/domain-model.md) — scenarios, scenario sets, audit runs, immutability invariants
-- [Deployment](docs/deployment.md) — Docker Compose, production hardening, GPU workers, upgrades
+- [Deployment](docs/deployment.md) — Docker Compose, production hardening, upgrades
 
 ## 🤝 Contributing
 

{simpleaudit_studio-0.2.2 → simpleaudit_studio-0.2.3}/README.md

@@ -25,7 +25,7 @@ A self-hostable platform for running reproducible AI model audits using the [Sim
 uvx simpleaudit-studio
 ```
 
-[`uvx`](https://docs.astral.sh/uv/#uvx) installs the [`simpleaudit-studio`](https://pypi.org/project/simpleaudit-studio/) package and runs it in an isolated environment. No Docker, no Postgres, no manual setup. Opens at http://localhost:8000 (login: `admin` / `admin12345`). A mock model server is pre-seeded so you can start exploring audit results immediately.
+[`uvx`](https://docs.astral.sh/uv/#uvx) installs the [`simpleaudit-studio`](https://pypi.org/project/simpleaudit-studio/) package and runs it in an isolated environment. No Docker, no Postgres, no manual setup. Opens at http://localhost:8000 (login: `studio` / `admin123`). A mock model server is pre-seeded so you can start exploring audit results immediately.
 
 🌐 Or skip the setup entirely — try the live demo: <a href="https://sushantgautam-simpleaudit-studio.hf.space" target="_blank" rel="noopener noreferrer">sushantgautam-simpleaudit-studio.hf.space</a>
 
@@ -36,7 +36,7 @@ The pre-seeded connections point at a built-in mock server. To run real audits,
 | Provider | Base URL | Example |
 |----------|----------|---------|
 | Ollama (local) | `http://localhost:11434/v1` | Qwen 3, Llama 3.3, Mistral |
-| vLLM (GPU) | `http://your-gpu-server:8000/v1` | Any HF model |
+| vLLM | `http://your-vllm-server:8000/v1` | Any HF model |
 | OpenAI | `https://api.openai.com/v1` | GPT-4o, o3 |
 | Together AI | `https://api.together.xyz/v1` | Many open models |
 | Groq | `https://api.groq.com/openai/v1` | Fast inference |
@@ -45,7 +45,7 @@ Set the API key in the connection form (leave blank for local servers that don't
 
 ## 🐳 Self-Hosting
 
-For teams, multi-user setups, or GPU worker pools, use Docker Compose:
+For teams or multi-user setups, use Docker Compose:
 
 ```bash
 git clone https://github.com/SushantGautam/SimpleAuditStudio
@@ -57,7 +57,7 @@ docker compose up -d
 
 Services: Web UI (:8000), PostgreSQL, Hatchet queue (:8888), Worker. Optional profiles: `--profile storage` (MinIO), `--profile mock` (mock model API).
 
-See [docs/deployment.md](docs/deployment.md) for production hardening, GPU workers, backups, and upgrades.
+See [docs/deployment.md](docs/deployment.md) for production hardening, backups, and upgrades.
 
 ## ✨ What You Can Do
 
@@ -84,7 +84,7 @@ Browser → Django/Gunicorn (:7860)
 
 - [Architecture](docs/architecture.md) — system design, service boundaries, deployment topology
 - [Domain Model](docs/domain-model.md) — scenarios, scenario sets, audit runs, immutability invariants
-- [Deployment](docs/deployment.md) — Docker Compose, production hardening, GPU workers, upgrades
+- [Deployment](docs/deployment.md) — Docker Compose, production hardening, upgrades
 
 ## 🤝 Contributing
 

{simpleaudit_studio-0.2.2 → simpleaudit_studio-0.2.3}/docs/architecture.md

@@ -9,7 +9,7 @@ The production platform must be:
 
 - self-hostable with Docker Compose
 - durable across web/API and worker restarts
-- safe for long-running and GPU-backed jobs
+- safe for long-running audit jobs
 - observable end to end
 - reproducible at the audit level
 - maintainable by multiple developers
@@ -39,10 +39,8 @@ flowchart TD
     Web --> Obj[(S3/MinIO Object Storage)]
     Web --> WF[Workflow/Queue System]
     WF --> CPU[CPU Workers]
-    WF --> GPU[GPU Workers]
     WF --> APIWorkers[External API Workers]
     CPU --> SA[SimpleAudit Engine]
-    GPU --> SA
     APIWorkers --> SA
     SA --> Obj
     SA --> PG
@@ -120,20 +118,16 @@ to credential material inside the worker environment.
 Workers register labels/capabilities:
 
 - `cpu`
-- `gpu`
-- `h200`
-- `gh200`
 - `external-api`
 - `local-inference`
 
-Jobs request required capabilities. The web server never needs GPU access.
+Jobs request required capabilities. The web server never needs direct model access.
 
 Concurrency limits are configured per pool:
 
 - global max concurrent audits
 - per-run max concurrent scenarios
 - per-model endpoint rate limit where known
-- GPU worker slot count
 
 ### 4.4 Idempotency
 
@@ -334,8 +328,7 @@ docker compose
 ├── postgres
 ├── minio
 ├── hatchet-server
-├── hatchet-worker-cpu
-├── hatchet-worker-gpu   # optional, same host or remote
+├── hatchet-worker
 ├── web
 └── collector            # optional
 ```

{simpleaudit_studio-0.2.2 → simpleaudit_studio-0.2.3}/docs/deployment.md

@@ -21,21 +21,22 @@ After startup, the user should get:
 - PostgreSQL initialized/migrated
 - object storage bucket initialized
 - workflow server available to workers
-- CPU worker running
-- optional GPU worker running if hardware/config present
+- worker running
 - health checks passing
 - documented first-run bootstrap command
 
 No undocumented manual database setup is allowed.
 
-First-run bootstrap:
+First-run bootstrap (idempotent, runs automatically on container start):
 
-1. create initial admin user from environment or management command
-2. create default project
-3. assign admin user to default project
+1. create the platform owner account (`BOOTSTRAP_USERNAME`, default `studio`) with `is_staff=True` — not a Django superuser
+2. create the Default workspace
+3. grant the owner an ADMIN membership in the Default workspace
 4. optionally import a starter scenario pack into that project
 5. print or log next steps for model endpoint registration
 
+Configuration via environment variables: `BOOTSTRAP_USERNAME`, `BOOTSTRAP_EMAIL`, `BOOTSTRAP_PASSWORD`, `BOOTSTRAP_PROJECT_NAME`.
+
 Bootstrap must be idempotent and must not require hand-editing SQL.
 
 ## 2. Canonical components
@@ -46,8 +47,7 @@ Bootstrap must be idempotent and must not require hand-editing SQL.
 | PostgreSQL | authoritative state | `postgres` |
 | MinIO/S3 | artifacts | `minio` |
 | Hatchet server | durable workflow/queue | `hatchet-server` |
-| CPU worker | runs audits on CPU/API models | `worker-cpu` |
-| GPU worker | runs local/GPU inference | `worker-gpu` optional |
+| Worker | runs audits (CPU + external API models) | `worker` |
 | OTel collector | traces/logs/metrics | optional |
 | Langfuse | LLM observability | optional |
 
@@ -81,7 +81,7 @@ Requirements:
 
 - isolated data
 - same migration path as production
-- access to representative GPU/API endpoints if available
+- access to representative model API endpoints if available
 
 ### 3.3 Production
 
@@ -345,8 +345,7 @@ Components:
 - PostgreSQL
 - MinIO/S3
 - Hatchet server
-- CPU worker
-- optional GPU worker
+- worker
 - web/API
 - optional OTel collector/Langfuse
 
@@ -358,7 +357,7 @@ Components:
 
 - PostgreSQL
 - Hatchet server
-- CPU worker
+- worker
 - web/API
 - local volume or managed object storage compatible with the same artifact interface
 
@@ -371,13 +370,11 @@ PostgreSQL, authentication, and object-storage-compatible artifact persistence.
 Single host:
 
 - scale worker concurrency within host resources
-- run GPU worker only if GPU available
 
 Multi host:
 
 - remote workers point to same Postgres/MinIO/Hatchet
 - add labels for capabilities
-- no web server access to GPU required
 - consider separate network segmentation for workers
 
 Do not scale horizontally until single-host bottlenecks are measured.

{simpleaudit_studio-0.2.2 → simpleaudit_studio-0.2.3}/pyproject.toml

@@ -4,7 +4,7 @@ build-backend = "hatchling.build"
 
 [project]
 name = "simpleaudit-studio"
-version = "0.2.2"
+version = "0.2.3"
 description = "Production AI model audit platform — one-liner demo mode via uvx"
 readme = "README.md"
 requires-python = ">=3.11"

{simpleaudit_studio-0.2.2 → simpleaudit_studio-0.2.3}/simpleaudit_studio/cli.py

@@ -100,7 +100,7 @@ def main() -> None:
     time.sleep(1)
 
     username = os.environ.get("BOOTSTRAP_USERNAME", "studio")
-    password = os.environ.get("BOOTSTRAP_PASSWORD", "admin12345")
+    password = os.environ.get("BOOTSTRAP_PASSWORD", "admin123")
 
     print("┌─────────────────────────────────────────────────────────┐")
     print("│                                                         │")
@@ -142,7 +142,7 @@ def _seed_demo_data() -> None:
     # Respect env vars (set by Dockerfile for HF Space, or defaults for local)
     username = os.environ.get("BOOTSTRAP_USERNAME", "studio")
     email = os.environ.get("BOOTSTRAP_EMAIL", "admin@localhost")
-    password = os.environ.get("BOOTSTRAP_PASSWORD", "admin12345")
+    password = os.environ.get("BOOTSTRAP_PASSWORD", "admin123")
     project_name = os.environ.get("BOOTSTRAP_PROJECT_NAME", "Demo Project")
 
     user, project = bootstrap_admin_and_default_project(