🧠 LLM Logger
LLM Logger is a Python logging framework designed for seamless integration with Large Language Models (LLMs).
It produces structured, LLM-friendly logs that can be easily summarized or reasoned about — even across massive, deeply nested codebases.
🚀 Features
- 🪵 Structured JSON logs (timestamped, contextual, machine-readable)
- 🧠 LLM-optimized format with
reason,inputs,outputs, and semantic tags - 📂 Log grouping (
start_group,end_group) and function spans - 🔌 Modular LLM backend support via environment variable:
Ollama(local models)Groq(LLama, Gemma via API)OpenAI(GPT-3.5 / GPT-4)
- 📊 Summarization CLI with smart filtering and token-aware chunking
- 💾 Cache for LLM calls with expiration/cleanup
- 🧪 Test harness to simulate deeply nested logs
📦 Installation
pip install ailogx
🛠️ Basic Usage
from ailogx.core import LLMLogger
log = LLMLogger("my-service")
log.llm_info("User login started", inputs={"username": "admin"})
log.llm_decision("Using 2FA", reason="high-risk user")
log.llm_error("Login failed", reason="Invalid OTP")
🔁 Function Span
with log.function_span("process_payment", reason="checkout flow"):
# your logic
pass
📂 Grouping Logs
log.start_group("req-42", reason="incoming API request")
# your logs here
log.end_group("req-42")
📊 LLM Summarization
🧠 Environment-based Backend Selection
Supports:
LLM_LOGGER_BACKEND=ollama(default)LLM_LOGGER_BACKEND=groqLLM_LOGGER_BACKEND=openai
🧾 Example
export LLM_LOGGER_BACKEND=groq # or 'openai', 'ollama'
export GROQ_API_KEY="your-groq-api-key"
- You must have a Groq account.
- Supported models: gemma3, llama3-70b, etc.
export OPENAI_API_KEY="your-openai-api-key"
- You must have an OpenAI API key.Models like gpt-3.5-turbo, gpt-4, etc. are supported.
For selecting Models :
| Backend | Env Var to Set | Example Value |
|----------|----------------|-----------------------|
| groq | GROQ_MODEL | llama-3-70b-8192 |
| openai | OPENAI_MODEL | gpt-4o |
| ollama | OLLAMA_MODEL | llama3 |
python -m ailogx.summarize simulated_logs/deep_nested_logs.jsonl --filter=smart --fast
For using with relevant intent:
python -m ailogx.summarize huge_app_logs.jsonl --filter=smart --fast --intent "focus on authentication and signup failures"
Or call from Python:
from ailogx.summarizer.summarizer import multi_pass_summarize
from ailogx.backends.registry import get_analyzer
import json
with open("llm_logs.jsonl") as f:
logs = [json.loads(line) for line in f]
summary = multi_pass_summarize(logs, get_analyzer())
print(summary)
🧪 Test Harness
Generate deep, nested logs for benchmarking:
python ailogx/core.py
Outputs:
llm_simulated_logs.jsonl(LLMLogger)standard_simulated_logs.log(Python logging)
🔁 Cache & Optimization
- ✅ LLM responses cached to
.cache/ - 🧠 Token-aware chunking
- 🔎 Smart filtering (
--filter=smart,--intent="auth errors") - ⚡
--fastmode for shallow summaries before full deep dives
Metadata
Release files for ailogx 1.1.0
For a detailed explanation of source distributions (sdists) and built distributions (wheels), please see the package formats documentation.
Source distribution (sdist)
| File | Size | Uploaded | Provenance | Actions |
|---|---|---|---|---|
| ailogx-1.1.0.tar.gz | 13.5 kB | Without provenance | Details for ailogx-1.1.0.tar.gz |
Built distribution (wheel)
| File | Interpreter | ABI | Platform | Actions Reset |
|---|---|---|---|---|
| ailogx-1.1.0-py3-none-any.whl | Python 3 | none | any | Details for ailogx-1.1.0-py3-none-any.whl |
Total release size: 31.3 kB
Release files / ailogx-1.1.0.tar.gz
| Download URL | ailogx-1.1.0.tar.gz |
|---|---|
| Size | 13.5 kB |
| Tags | Source |
|
SHA-256 checksum How to use checksums |
ca3a08630e82bb65ee1aefe094066ff7d8f8a564fc05e168cf34946ed4aa14fd
|
|
BLAKE2b-256 checksum How to use checksums |
8d7c7435e36c2c11432f867be606e71ee02fe54d8cb6fd9c51694427f54fa8ca
|
| Upload date | |
|
Uploaded using Trusted Publishing? What is trusted publishing? |
No |
| Uploaded via |
twine/6.1.0 CPython/3.13.5
|
Release files / ailogx-1.1.0-py3-none-any.whl
| Download URL | ailogx-1.1.0-py3-none-any.whl |
|---|---|
| Size | 17.8 kB |
| Tags | Python 3 |
|
SHA-256 checksum How to use checksums |
88ad3293740be129c0eb7628bd0c2422c3961f0e14a4cf5e27d8d07eb1f129ac
|
|
BLAKE2b-256 checksum How to use checksums |
a5ad2f98189a8c444e047dd767d6ad0116c0f20e885d84b5bc2484d34f275629
|
| Upload date | |
|
Uploaded using Trusted Publishing? What is trusted publishing? |
No |
| Uploaded via |
twine/6.1.0 CPython/3.13.5
|
Security for ailogx 1.1.0
Provenance
Provenance describes where a file came from. On PyPI, provenance is shared via attestations, which provide a verifiable record of the build or publishing details. View details, limitations and caveats.
No provenance found
None of the files in this release were uploaded with provenance metadata.
A note on package security
PyPI is a repository for distributing packages, not a service for auditing them. While third-party security researchers and our staff do their best to detect and remove malicious packages, PyPI cannot guarantee the quality, security or suitability of any given package.
To learn more about protecting yourself from malicious actors, visit our package security documentation.