epi-recorder 1.0.0__py3-none-any.whl

epi_recorder-1.0.0.dist-info/METADATA

@@ -0,0 +1,503 @@
+Metadata-Version: 2.4
+Name: epi-recorder
+Version: 1.0.0
+Summary: Evidence Packaged Infrastructure for AI workflows - Record, Verify, and Replay AI runs
+Author-email: EPI Project <contact@epi.dev>
+Maintainer-email: EPI Project <contact@epi.dev>
+License: Apache-2.0
+Project-URL: Homepage, https://github.com/epi-project/epi-recorder
+Project-URL: Documentation, https://epi.dev/docs
+Project-URL: Repository, https://github.com/epi-project/epi-recorder
+Project-URL: Issues, https://github.com/epi-project/epi-recorder/issues
+Keywords: ai,reproducibility,verification,llm,evidence,openai,cryptography,workflow,audit
+Classifier: Development Status :: 4 - Beta
+Classifier: Intended Audience :: Developers
+Classifier: Intended Audience :: Science/Research
+Classifier: License :: OSI Approved :: Apache Software License
+Classifier: Operating System :: OS Independent
+Classifier: Programming Language :: Python :: 3
+Classifier: Programming Language :: Python :: 3.11
+Classifier: Programming Language :: Python :: 3.12
+Classifier: Topic :: Scientific/Engineering :: Artificial Intelligence
+Classifier: Topic :: Security :: Cryptography
+Classifier: Topic :: Software Development :: Libraries :: Python Modules
+Requires-Python: >=3.11
+Description-Content-Type: text/markdown
+License-File: LICENSE
+Requires-Dist: pydantic>=2.0.0
+Requires-Dist: cryptography>=41.0.0
+Requires-Dist: cbor2>=5.6.0
+Requires-Dist: typer[all]>=0.12.0
+Requires-Dist: rich>=13.0.0
+Provides-Extra: dev
+Requires-Dist: pytest>=8.0.0; extra == "dev"
+Requires-Dist: pytest-asyncio>=0.23.0; extra == "dev"
+Requires-Dist: pytest-cov>=4.1.0; extra == "dev"
+Requires-Dist: black>=24.0.0; extra == "dev"
+Requires-Dist: ruff>=0.3.0; extra == "dev"
+Dynamic: license-file
+
+# EPI - Evidence Packaged Infrastructure
+
+**The "PDF for AI Workflows"** — Self-contained, cryptographically verified evidence packages for AI systems.
+
+[![License](https://img.shields.io/badge/license-Apache%202.0-blue.svg)](LICENSE)
+[![Python](https://img.shields.io/badge/python-3.11+-blue.svg)](https://www.python.org/downloads/)
+[![Status](https://img.shields.io/badge/status-MVP-green.svg)]()
+
+---
+
+## 🚀 Quick Start
+
+### **Python API** (Recommended for Developers)
+
+```python
+from epi_recorder import record
+
+# Wrap your AI code with a context manager
+with record("my_workflow.epi", workflow_name="Demo"):
+    # Your AI code runs normally - automatically recorded!
+    response = openai.chat.completions.create(
+        model="gpt-4",
+        messages=[{"role": "user", "content": "Hello!"}]
+    )
+    
+    # Optionally log custom events
+    epi.log_step("calculation", {"result": 42})
+
+# .epi file is automatically created, signed, and ready to verify!
+```
+
+### **CLI** (For Shell Scripts & CI/CD)
+
+```bash
+# Install
+pip install epi-recorder
+
+# Record a workflow
+epi record --out demo.epi -- python my_ai_script.py
+
+# Verify integrity and authenticity
+epi verify demo.epi
+
+# View in browser
+epi view demo.epi
+```
+
+**That's it!** Your AI workflow is now captured, signed, and viewable.
+
+---
+
+## 🎯 What is EPI?
+
+EPI captures **everything** that happens during an AI workflow:
+- 🤖 **LLM API calls** (prompts, responses, tokens, latency)
+- 🔒 **Secrets redacted** automatically (15+ patterns)
+- 📦 **Files and artifacts** (content-addressed)
+- 🖥️ **Environment snapshot** (OS, Python, packages)
+- ✅ **Cryptographically signed** (Ed25519)
+- 📊 **Beautiful timeline viewer** (static HTML)
+
+All packaged into a **single `.epi` file** that anyone can verify and replay.
+
+---
+
+## 🌟 Why EPI?
+
+### **The Problem**
+
+❌ 70% of AI research cannot be reproduced  
+❌ AI models fail mysteriously in production  
+❌ Cannot prove how AI decisions were made  
+❌ "It worked on my machine" debugging nightmare  
+
+### **The Solution**
+
+✅ **Record**: Capture complete AI workflows with one command  
+✅ **Verify**: Cryptographic proof of authenticity  
+✅ **Share**: Single file contains everything  
+✅ **Replay**: Deterministic reproduction (offline mode)  
+✅ **Audit**: Full transparency for compliance  
+
+---
+
+## 📖 Core Features
+
+### 🎬 **Recording**
+```bash
+epi record --out experiment.epi -- python train.py
+```
+Automatically captures:
+- OpenAI API calls (GPT-4, GPT-3.5, etc.)
+- Shell commands and outputs
+- Python execution context
+- Generated files and artifacts
+- Environment variables (redacted)
+
+### 🔐 **Security by Default**
+- **Auto-redacts secrets**: API keys, tokens, credentials
+- **Ed25519 signatures**: Cryptographic proof of authenticity
+- **Frictionless**: Auto-generates keypair on first run
+- **No secret leakage**: 15+ regex patterns protect sensitive data
+
+### ✅ **Verification**
+```bash
+epi verify experiment.epi
+```
+Three-level verification:
+1. **Structural**: Valid ZIP format and schema
+2. **Integrity**: SHA-256 file hashes match
+3. **Authenticity**: Ed25519 signature valid
+
+### 👁️ **Beautiful Viewer**
+```bash
+epi view experiment.epi
+```
+Opens in your browser with:
+- Interactive timeline of all steps
+- LLM chat bubbles (prompts & responses)
+- Trust badges (signed/unsigned)
+- Artifact previews
+- **Zero code execution** (pure JSON rendering)
+
+---
+
+## 🎓 Use Cases
+
+### **AI Researchers**
+```bash
+# Submit verifiable research
+epi record --out paper_experiment.epi -- python reproduce.py
+```
+✅ 100% reproducible methodology  
+✅ Eliminates "it worked on my machine"  
+✅ Speeds up peer review  
+
+### **Enterprise AI Teams**
+```bash
+# Capture production AI runs
+epi record --out prod_run.epi -- python deploy_model.py
+```
+✅ Audit trails for compliance (EU AI Act, SOC 2)  
+✅ Debug production failures instantly  
+✅ Version control for AI systems  
+
+### **Software Engineers**
+```bash
+# Perfect bug reproduction
+epi record --out bug_report.epi -- python flaky_test.py
+```
+✅ Share exact failing conditions  
+✅ Debug AI features faster  
+✅ Stable CI/CD for AI features  
+
+---
+
+## 🛠️ Installation
+
+### **From Source** (Development)
+```bash
+cd epi-recorder
+pip install -e .
+```
+
+### **Requirements**
+- Python 3.11+
+- Windows, macOS, or Linux
+
+---
+
+## 📚 Commands
+
+### **`epi record`**
+Record a workflow into a `.epi` file.
+
+```bash
+epi record --out run.epi -- python script.py [args...]
+
+Options:
+  --out PATH              Output .epi file (required)
+  --no-sign              Don't sign the manifest
+  --no-redact            Disable secret redaction
+```
+
+### **`epi verify`**
+Verify `.epi` file integrity and authenticity.
+
+```bash
+epi verify run.epi
+
+Options:
+  --json        Output as JSON
+  --verbose     Verbose output
+```
+
+### **`epi view`**
+Open `.epi` file in browser.
+
+```bash
+epi view run.epi
+```
+
+### **`epi keys`**
+Manage Ed25519 key pairs.
+
+```bash
+epi keys generate --name mykey
+epi keys list
+epi keys export --name mykey
+```
+
+---
+
+## 🐍 Python API
+
+### **Why Use the Python API?**
+
+The Python API is the **recommended way** to integrate EPI into your AI applications:
+- ✅ **Zero CLI overhead** - no shell commands needed
+- ✅ **Seamless integration** - just wrap your code with `with record()`
+- ✅ **Auto-captures OpenAI** - automatically records all LLM calls
+- ✅ **Custom logging** - manually log steps and artifacts
+- ✅ **Auto-signed** - cryptographic signatures by default
+
+### **Basic Usage**
+
+```python
+from epi_recorder import record
+
+with record("experiment.epi", workflow_name="My Experiment"):
+    # Your code here - automatically recorded
+    result = train_model()
+    print(f"Result: {result}")
+```
+
+### **With Custom Logging**
+
+```python
+from epi_recorder import record
+from pathlib import Path
+
+with record("workflow.epi", workflow_name="Data Processing", tags=["v1.0", "prod"]) as epi:
+    # Process data
+    data = load_data("input.csv")
+    
+    # Log custom steps
+    epi.log_step("data.loaded", {
+        "rows": len(data),
+        "columns": list(data.columns)
+    })
+    
+    # Process...
+    results = process(data)
+    
+    # Save output
+    results.to_csv("output.csv")
+    
+    # Capture the output file
+    epi.log_artifact(Path("output.csv"))
+    
+    # Log summary
+    epi.log_step("processing.complete", {
+        "status": "success",
+        "output_rows": len(results)
+    })
+```
+
+### **With OpenAI (Auto-Recorded)**
+
+```python
+from epi_recorder import record
+import openai
+
+with record("chat_session.epi", workflow_name="Customer Support"):
+    # OpenAI calls are automatically captured!
+    response = openai.chat.completions.create(
+        model="gpt-4",
+        messages=[
+            {"role": "system", "content": "You are a helpful assistant."},
+            {"role": "user", "content": "What is quantum computing?"}
+        ]
+    )
+    
+    print(response.choices[0].message.content)
+    
+    # API keys are automatically redacted in the recording!
+```
+
+### **Advanced Options**
+
+```python
+from epi_recorder import EpiRecorderSession
+
+with EpiRecorderSession(
+    "advanced.epi",
+    workflow_name="Advanced Workflow",
+    tags=["experiment", "2024-10-29"],
+    auto_sign=True,        # Sign with default key (default: True)
+    redact=True,          # Redact secrets (default: True)
+    default_key_name="my-key"  # Custom signing key
+) as epi:
+    # Your workflow
+    pass
+```
+
+### **Manual LLM Logging** (For Custom Integrations)
+
+```python
+from epi_recorder import record
+
+with record("custom_llm.epi") as epi:
+    # For custom/proprietary LLM APIs
+    epi.log_llm_request("claude-3-opus", {
+        "messages": [{"role": "user", "content": "Hello"}],
+        "temperature": 0.7
+    })
+    
+    # ... make actual API call ...
+    
+    epi.log_llm_response({
+        "model": "claude-3-opus",
+        "content": "Hello! How can I help?",
+        "tokens": 25
+    })
+```
+
+### **Error Handling**
+
+Recordings are saved even if errors occur:
+
+```python
+from epi_recorder import record
+
+try:
+    with record("debug_session.epi", workflow_name="Debug") as epi:
+        epi.log_step("process.start", {"status": "ok"})
+        
+        # Code that might fail
+        result = risky_operation()
+        
+except Exception as e:
+    print(f"Error: {e}")
+    # .epi file is still created with error logs!
+```
+
+### **Running the Examples**
+
+```bash
+# See examples directory
+python examples/api_example.py
+
+# Verify the generated .epi files
+epi verify example_basic.epi
+
+# View in browser
+epi view example_basic.epi
+```
+
+---
+
+## 🔒 Security
+
+### **Automatic Redaction**
+EPI automatically removes:
+- OpenAI API keys (`sk-...`)
+- Anthropic API keys (`sk-ant-...`)
+- AWS credentials (`AKIA...`)
+- GitHub tokens (`ghp_...`)
+- Bearer tokens, JWT tokens
+- Database connection strings
+- Private keys (PEM format)
+
+### **Cryptographic Signing**
+- **Algorithm**: Ed25519 (RFC 8032)
+- **Key Size**: 256 bits
+- **Hash**: SHA-256 with canonical CBOR
+- **Storage**: `~/.epi/keys/` (secure permissions)
+
+---
+
+## 🧪 Example
+
+```python
+# chat_example.py
+from openai import OpenAI
+
+client = OpenAI()
+
+response = client.chat.completions.create(
+    model="gpt-3.5-turbo",
+    messages=[
+        {"role": "user", "content": "Explain quantum computing"}
+    ]
+)
+
+print(response.choices[0].message.content)
+```
+
+```bash
+# Record it
+epi record --out chat.epi -- python chat_example.py
+
+# Verify it
+epi verify chat.epi
+# ✅ Trust Level: HIGH
+
+# View it
+epi view chat.epi
+# Opens timeline in browser
+```
+
+---
+
+## 🧑‍💻 Development
+
+### **Running Tests**
+```bash
+pytest tests/ -v --cov=epi_core --cov=epi_cli
+```
+
+### **Project Structure**
+```
+epi-recorder/
+├── epi_core/           # Core logic
+├── epi_cli/            # CLI commands
+├── epi_recorder/       # Runtime capture
+├── epi_viewer_static/  # Static viewer
+├── tests/              # Test suite
+└── docs/               # Specification
+```
+
+---
+
+## 📊 Status
+
+✅ **Phase 0**: Foundation (complete)  
+✅ **Phase 1**: Trust Layer (complete)  
+✅ **Phase 2**: Recorder MVP (complete)  
+✅ **Phase 3**: Viewer MVP (complete)  
+🔄 **Phase 4**: Polish & docs (in progress)  
+
+**Current**: Keystone MVP - Production Ready
+
+---
+
+## 📄 License
+
+Apache License 2.0
+
+---
+
+## 🙏 Acknowledgments
+
+Built with:
+- [Pydantic](https://pydantic.dev/)
+- [Typer](https://typer.tiangolo.com/)
+- [Rich](https://rich.readthedocs.io/)
+- [cryptography](https://cryptography.io/)
+
+---
+
+**Made with ❤️ for the AI community**
+
+*Turning opaque AI runs into transparent, portable digital proofs.*

epi_recorder-1.0.0.dist-info/RECORD

@@ -0,0 +1,25 @@
+epi_cli/__init__.py,sha256=SSpuDlDqVBbPvR21CCRcNOB61EyB8jAUzNd8XISADa8,98
+epi_cli/keys.py,sha256=0ytG_rrwbn3YA_girZb_BNG-2HLnbx9WIKHwdDcfmHM,9116
+epi_cli/main.py,sha256=5rhqkmsu6Cy_mOkU7wYmrhhmfZors6P6JhlWd6KcNqY,3452
+epi_cli/record.py,sha256=2YLMDMojEe8uH7t3yHUtLfL1uAV9uEbAM4IsUoaoCF8,6975
+epi_cli/verify.py,sha256=wg3K41rY0dmKpLfEfzWl1_J8twThz8wrL7BG74XkzSQ,8210
+epi_cli/view.py,sha256=MJebDImddv-P0nb3-Mch1R5528gjvbpLxEsu5zxZqeI,2470
+epi_core/__init__.py,sha256=LD57Jbx5gsxxfntPoUwYhDd5EgEyyDLZr-0qA1BQJBc,312
+epi_core/container.py,sha256=nO0Z4pdvTwgKaPk7c6NmRfwGoesNOj6GOK5oabvIh3k,11949
+epi_core/redactor.py,sha256=YBA8of5ZSsOYAjJOlveDyMgNuFRUIUu_8dDs5vuzsNY,9273
+epi_core/schemas.py,sha256=hqX6tp2AiAuyNLz3MylYjZX5d7uNUYD-liZ5k7yzCM8,3487
+epi_core/serialize.py,sha256=rx6tEnfXX1gW-AGEUpVWPEB9zsaIZ4qYn65s-7aQQtA,4736
+epi_core/trust.py,sha256=yk-ZrqVl9nTabAdSOfCTNUbAWSJvMds7wxc0ghdtEPM,7551
+epi_recorder/__init__.py,sha256=WDgzWMiTJVnA26sXaF5Id8ptQawLlkLeU2pVy268lQU,405
+epi_recorder/api.py,sha256=opVfDb0wjUXtVTRz1CFD7F-ZngWrzfo0TWga8C3zhlE,13952
+epi_recorder/bootstrap.py,sha256=yE1y0rjQ-1D-ZhWFipX_6qXFTA2BPlEm8Ft4Z0iex88,1845
+epi_recorder/environment.py,sha256=DAudeqABQQjKLgxsuN23klJQMrdjnLuj4R3SOr2AjDA,5990
+epi_recorder/patcher.py,sha256=Cr6orXgxGWiUlasekyTf1EUXjDsQK86WqqQ0nXVQyaw,11817
+epi_recorder-1.0.0.dist-info/licenses/LICENSE,sha256=aUgjAzBUkmKIZy3AZ_XAZWoh3kzkRlkOc3JHLVfyF-A,11540
+epi_viewer_static/app.js,sha256=3yZpMZtMDl6Ug36jB4R_ZZkYlt3WCFX0LNmYZIAZ_DU,9662
+epi_viewer_static/index.html,sha256=OMEggB4u4O9eIDAO2wb9u_qOwZwxcLl4K9BjcK6vaZg,3136
+epi_recorder-1.0.0.dist-info/METADATA,sha256=0WTyAo-jZdVNf3G1qbYMM4GGWMUN6BhZv6WO_mN85Yo,12826
+epi_recorder-1.0.0.dist-info/WHEEL,sha256=_zCd3N1l69ArxyTb8rzEoP9TpbYXkqRFSNOD5OuxnTs,91
+epi_recorder-1.0.0.dist-info/entry_points.txt,sha256=MfMwqVRx_yMGbuPpiyjz2f8fQp8TUbHmRC1H_bupoyM,41
+epi_recorder-1.0.0.dist-info/top_level.txt,sha256=nKpWVQUoGWK-gyLLVbrxibQUlIzI0TowNzMDondlHpI,48
+epi_recorder-1.0.0.dist-info/RECORD,,

epi_recorder-1.0.0.dist-info/WHEEL

@@ -0,0 +1,5 @@
+Wheel-Version: 1.0
+Generator: setuptools (80.9.0)
+Root-Is-Purelib: true
+Tag: py3-none-any
+

epi_recorder-1.0.0.dist-info/entry_points.txt

@@ -0,0 +1,2 @@
+[console_scripts]
+epi = epi_cli.main:app