agentkavach

v0.1.0 suspicious
4.0
Medium Risk

Hard budget limits and guardrails for LLM APIs (OpenAI, Anthropic, Google, Mistral).

🤖 AI Analysis

Final verdict: SUSPICIOUS

The package shows signs of potential risk due to its metadata and lack of detailed repository and maintainer history, despite having low scores for direct security risks like shell execution and credential harvesting.

  • Metadata risk score of 6 out of 10
  • Lack of repository and maintainer history
Per-check LLM notes
  • Network: The observed network calls are typical for making HTTP requests and could be part of normal functionality, but the absence of proper input validation or encryption raises some concern.
  • Shell: No shell execution patterns were detected.
  • Obfuscation: No obfuscation patterns detected, indicating low risk of malicious activity.
  • Credentials: No credential harvesting patterns detected, indicating low risk of malicious activity.
  • Metadata: The package is suspicious due to lack of repository and maintainer history, indicating potential risk.

🔬 Heuristic Checks

⚠ Outbound Network Calls score 9.0

Found 6 network call pattern(s)

  • ncode() req = urllib.request.Request( url, data=d
  • ) with urllib.request.urlopen(req, timeout=10) as resp: logger
  • } self._client = httpx.Client(timeout=_INGEST_TIMEOUT_SECONDS) self._buffer = buff
  • ate_id self._client = httpx.Client(timeout=_TIMEOUT_SECONDS) def send( self,
  • ng_key self._client = httpx.Client(timeout=_TIMEOUT_SECONDS) def send( self,
  • ok_url self._client = httpx.Client(timeout=_TIMEOUT_SECONDS) def send( self,
✓ Code Obfuscation

No obfuscation patterns detected

✓ Shell / Subprocess Execution

No shell execution patterns detected

✓ Credential Harvesting

No credential harvesting patterns detected

✓ Typosquatting

No typosquatting candidates detected

✓ Registered Email Domain

Email domain looks legitimate: gmail.com>

✓ Suspicious Page Links

All external links appear legitimate

⚠ Git Repository History score 3.0

Repository not found (deleted or private)

  • Repository not found (deleted or private)
⚠ Maintainer History score 6.0

3 maintainer concern(s) found

  • Only one version has ever been released — brand new package
  • Author name is missing or very short
  • Author "" appears to have only 1 package on PyPI (new or inactive account)
✓ Known CVE Vulnerabilities

No known vulnerabilities found in OSV database.

💡 AI App Starter Prompt

Use this prompt to build a project with agentkavach 
Develop a mini-application called 'GuardedGenie' that leverages the 'agentkavach' Python package to manage and control the usage of large language model (LLM) APIs such as OpenAI, Anthropic, Google, and Mistral. This application will serve as a versatile tool for developers and users who want to ensure their interactions with these powerful models stay within specified budget limits and adhere to set guardrails. Here’s a detailed breakdown of the application's requirements and functionalities:

1. **User Interface**: Design a simple yet intuitive command-line interface (CLI) where users can interact with the application. Users should be able to select which LLM API they wish to use, set up budget limits, and define guardrails for their queries.

2. **Budget Management**: Implement a feature that allows users to specify a monthly or per-session budget limit for API calls. The application should track the cost of each call and alert the user when they are approaching their budget limit.

3. **Guardrails Setup**: Users should have the ability to set up guardrails that define the types of queries they allow or disallow. For example, they could specify that no queries containing certain keywords or topics should be processed.

4. **Integration with LLM APIs**: Ensure seamless integration with at least three different LLM APIs (e.g., OpenAI, Anthropic, and Google). Each integration should respect the budget limits and guardrails defined by the user.

5. **Logging and Reporting**: Implement logging functionality to record all API calls, including the query, response, timestamp, and cost. Provide a reporting feature that summarizes the usage over a given period, highlighting any breaches of budget limits or guardrails.

6. **Customizability**: Allow users to customize the behavior of 'GuardedGenie' by configuring settings through a configuration file or directly from the CLI.

7. **Error Handling**: Ensure robust error handling mechanisms are in place to gracefully handle situations like exceeding budget limits, violating guardrails, or encountering API errors.

8. **Security**: Since the application will handle sensitive information such as API keys and budgets, ensure proper security measures are taken to protect this data.

The 'agentkavach' package will play a crucial role in enforcing the budget limits and guardrails across all integrated LLM APIs. Specifically, you'll utilize its capabilities to monitor API usage, enforce spending limits, and validate queries against defined guardrails. Your task is to build a fully functional 'GuardedGenie' application that not only showcases the power of 'agentkavach' but also provides real-world value to developers and end-users looking to responsibly leverage LLM APIs.