v0.1.4

NeuralWatch v0.1.4

This release adds PyPI documentation links and status badges, along with .gitignore cleanups.

📦 SDK Installation

pip install neuralwatch-splunk

📝 Changelog (since v0.1.3)

Added

• Added a live PyPI version status badge to the top header of README.md and README_PYPI.md.
• Added direct installation page references pointing back to pypi.org/project/neuralwatch-splunk.

Changed

• Refactored .gitignore pattern exclusions to keep the developer workspace clean of build artifacts and environment configurations.

🛠️ Quick Start

import openai
from neuralwatch_sdk import instrument

instrument(
    service="my-service",
    team="my-team",
    splunk_hec_url="https://your-splunk-hec-url:8088/services/collector/event",
    splunk_hec_token="your-token"
)

# Use your OpenAI / Anthropic clients normally!

v0.1.3

NeuralWatch v0.1.3

This release introduces layout modernizations, visual diagrams, and optimizations for PyPI compatibility.

📦 SDK Installation

pip install neuralwatch-splunk

📝 Changelog (since v0.1.1)

Added

• Added an official root-level architecture diagram architecture_diagram.md highlighting:
  • Application-to-Splunk communication paths.
  • AI Model and Agent integration layers.
  • End-to-end data sequence flow.
• Integrated modern mermaid flowcharts to visually display the system metrics architecture, Prompt Injection Sentinel pipeline, and EU AI Act Compliance scoring engine.
• Added a dedicated PyPI-optimized package documentation file (README_PYPI.md) to resolve limited Markdown rendering capabilities on the public index pages.

Fixed

• Fixed YAML syntax validation errors in the GitHub Actions publish workflow by moving token/OIDC authorization blocks to the top-level permissions section.
• Fixed typos, normalized em-dashes into standard hyphens across documentation, and clarified import namespaces vs. installation instructions in the Quick Start guide.

🛠️ Quick Start

import openai
from neuralwatch_sdk import instrument

instrument(
    service="my-service",
    team="my-team",
    splunk_hec_url="https://your-splunk-hec-url:8088/services/collector/event",
    splunk_hec_token="your-token"
)

# Use your OpenAI / Anthropic clients normally!

v0.1.2

NeuralWatch v0.1.2 - Latest Release

We are excited to release the initial version of NeuralWatch (neuralwatch-splunk), a zero-code-change AI observability and compliance tracking platform for Splunk.

🚀 Key Features

• Zero-Code-Change Instrumentation: Automatically patches OpenAI and Anthropic Python SDKs with a single instrument() call.
• Asynchronous Telemetry Ingest: Streams latency, token usage, cost metrics, and error rates via a non-blocking queue directly to Splunk HEC.
• Prompt Injection Sentinel: Classifies prompts in real time using a heuristic Foundation-Sec-inspired classifier (LOW, MEDIUM, HIGH, CRITICAL risk).
• EU AI Act Compliance Scoring: Real-time metrics mapping telemetry data to regulatory articles (Articles 9, 13, 14, 17, and 72) in Splunk.
• Natural Language Splunk Query Agent (MCP): Translates plain English queries into SPL and runs searches against telemetry indexes.
• Pre-packaged Splunk App: Visualizations and dashboards out-of-the-box (splunk_app.tar.gz).
• Trusted Publishing: Configured GitHub Actions OIDC integration for secure, automated releases to PyPI.

📦 SDK Installation

pip install neuralwatch-splunk

v0.1.1

NeuralWatch v0.1.1 - Initial Release

We are excited to release the initial version of NeuralWatch (neuralwatch-splunk), a zero-code-change AI observability and compliance tracking platform for Splunk.

🚀 Key Features

• Zero-Code-Change Instrumentation: Automatically patches OpenAI and Anthropic Python SDKs with a single instrument() call.
• Asynchronous Telemetry Ingest: Streams latency, token usage, cost metrics, and error rates via a non-blocking queue directly to Splunk HEC.
• Prompt Injection Sentinel: Classifies prompts in real time using a heuristic Foundation-Sec-inspired classifier (LOW, MEDIUM, HIGH, CRITICAL risk).
• EU AI Act Compliance Scoring: Real-time metrics mapping telemetry data to regulatory articles (Articles 9, 13, 14, 17, and 72) in Splunk.
• Natural Language Splunk Query Agent (MCP): Translates plain English queries into SPL and runs searches against telemetry indexes.
• Pre-packaged Splunk App: Visualizations and dashboards out-of-the-box (splunk_app.tar.gz).
• Trusted Publishing: Configured GitHub Actions OIDC integration for secure, automated releases to PyPI.

📦 SDK Installation

pip install neuralwatch-splunk

Quick Start:

import openai
from neuralwatch_splunk import instrument

instrument(
    service="my-service",
    team="my-team",
    splunk_hec_url="https://your-splunk-hec-url:8088/services/collector/event",
    splunk_hec_token="your-token"
)

# Use your OpenAI / Anthropic clients normally!

v0.1.0

NeuralWatch v0.1.0 - Initial Release

We are excited to release the initial version of NeuralWatch, a zero-code-change AI observability and compliance tracking platform for Splunk.

🚀 Key Features

• Zero-Code-Change Instrumentation: Automatically patches OpenAI and Anthropic Python SDKs with a single instrument() call.
• Asynchronous Telemetry Ingest: Streams latency, token usage, cost metrics, and error rates via a non-blocking queue directly to Splunk HEC.
• Prompt Injection Sentinel: Classifies prompts in real time using a heuristic Foundation-Sec-inspired classifier (LOW, MEDIUM, HIGH, CRITICAL risk).
• EU AI Act Compliance Scoring: Real-time metrics mapping telemetry data to regulatory articles (Articles 9, 13, 14, 17, and 72) in Splunk.
• Natural Language Splunk Query Agent (MCP): Translates plain English queries into SPL and runs searches against telemetry indexes.
• Pre-packaged Splunk App: Visualizations and dashboards out-of-the-box (splunk_app.tar.gz).

📦 SDK Installation

pip install neuralwatch-splunk

Quick Start:

import openai
from neuralwatch_splunk import instrument

instrument(
    service="my-service",
    team="my-team",
    splunk_hec_url="https://your-splunk-hec-url:8088/services/collector/event",
    splunk_hec_token="your-token"
)

# Use your OpenAI / Anthropic clients normally!