Agentic Bot

Agentic Bot

Last updated: 2025-11-15 18:49:11

agentic-bot.py

Auto-generated from .github/scripts/agentic-bot.py

Agentic PR Bot API Documentation

Overview

The Agentic PR Bot is a Python module designed to respond to action requests in PR comments and make code changes. It utilizes a large language model (LLM) to generate code changes based on user requests. The module is intended for use in GitHub Actions workflows.

Exports

The following functions are exported by the module:

• is_authorized(user): Checks if a user is authorized to trigger actions.
• classify_intent(comment_text): Classifies the intent of a PR comment.
• load_changed_files(): Loads the changed files in a PR.
• generate_code_changes(action_description, files): Generates code changes using an LLM.
• parse_and_apply_changes(llm_output, original_files): Parses LLM output and applies changes to files.
• commit_changes(files_changed, action_description): Commits changes to a repository.

Usage Examples

Checking Authorization

if is_authorized("github_username"):
    print("User is authorized")
else:
    print("User is not authorized")

Classifying Intent

intent = classify_intent("Please add comments to file.ts")
print(intent["category"])  # Output: CLEAR_ACTION
print(intent["action"])  # Output: add comments to file.ts

Loading Changed Files

changed_files = load_changed_files()
print(changed_files)  # Output: ["file1.ts", "file2.js"]

Generating Code Changes

llm_output = generate_code_changes("add comments to file.ts", ["file1.ts", "file2.js"])
print(llm_output)  # Output: LLM output containing modified file contents

Parsing and Applying Changes

original_files = {"file1.ts": "original content"}
files_modified = parse_and_apply_changes(llm_output, original_files)
print(files_modified)  # Output: ["file1.ts"]

Committing Changes

commit_hash = commit_changes(["file1.ts"], "add comments to file.ts")
print(commit_hash)  # Output: commit hash

Parameters

• is_authorized(user): user (str) - The GitHub username to check for authorization.
• classify_intent(comment_text): comment_text (str) - The PR comment text to classify.
• load_changed_files(): No parameters.
• generate_code_changes(action_description, files):
  • action_description (str) - The description of the action to generate code changes for.
  • files (list[str]) - The list of files to generate code changes for.
• parse_and_apply_changes(llm_output, original_files):
  • llm_output (str) - The LLM output containing modified file contents.
  • original_files (dict[str, str]) - The original file contents for security checking.
• commit_changes(files_changed, action_description):
  • files_changed (list[str]) - The list of files that have been modified.
  • action_description (str) - The description of the action that was performed.

Return Values

• is_authorized(user): bool - Whether the user is authorized to trigger actions.
• classify_intent(comment_text): dict - A dictionary containing the classified intent, including the category, action, confidence, and reasoning.
• load_changed_files(): list[str] - The list of changed files in the PR.
• generate_code_changes(action_description, files): str - The LLM output containing modified file contents.
• parse_and_apply_changes(llm_output, original_files): list[str] - The list of files that have been modified.
• commit_changes(files_changed, action_description): str - The commit hash of the committed changes, or None if the commit failed.