0Pricing
AI Agents · Lesson

Commit History and Diff Analysis

Walking commit history, extracting changed files, and summarizing diffs.

Commit History and Diff Analysis is a free AI Agents lesson on CoddyKit — lesson 4 of 4. You can read the complete lesson below for free — then practise it hands-on in the browser with a built-in code editor and a 24/7 AI tutor. It is part of the AI Agents learning path, one of 4 lessons in the course, and your progress syncs across the web and the CoddyKit app.

Why Analyze Commit History?

Commit history is a gold mine for agents: it reveals what changed, when, and by whom. Agents use commit history for changelog generation, code review summaries, developer activity reports, detecting risk (large changes close to releases), and understanding the evolution of a codebase.

from github import Github
import os

g = Github(token=os.environ['GITHUB_TOKEN'])
repo = g.get_repo('myorg/myrepo')

# Get recent commits on the default branch
commits = repo.get_commits()
print(f'Total commits: {commits.totalCount}')

# Preview the last 5
for commit in commits[:5]:
    print(f'{commit.sha[:8]} | {commit.commit.author.name}')
    print(f'  {commit.commit.message.split(chr(10))[0][:60]}')
    print(f'  {commit.commit.author.date.date()}')

Filtering Commits by Time Range

Use since and until parameters to get commits in a specific date range. Both take Python datetime objects. This is the primary tool for generating weekly/monthly activity reports.

import datetime
from github import Github
import os

g = Github(token=os.environ['GITHUB_TOKEN'])
repo = g.get_repo('myorg/myrepo')

# Last 7 days of commits
now = datetime.datetime.utcnow()
last_week = now - datetime.timedelta(days=7)

recent_commits = repo.get_commits(
    since=last_week,
    until=now
)

print(f'Commits in last 7 days: {recent_commits.totalCount}')

# By specific author
author_commits = repo.get_commits(
    since=last_week,
    author='alice'  # GitHub username
)
print(f'Alice\'s commits last week: {author_commits.totalCount}')

# On a specific branch
branch_commits = repo.get_commits(
    sha='feature/new-api',
    since=last_week
)

Commit Stats: Additions and Deletions

Each commit object has a stats property with additions, deletions, and total change counts across all changed files. This gives a quick sense of the size of a change.

from github import Github
import os

g = Github(token=os.environ['GITHUB_TOKEN'])
repo = g.get_repo('myorg/myrepo')

commit = repo.get_commit('abc12345sha')

# Overall stats
print(f'SHA: {commit.sha[:8]}')
print(f'Message: {commit.commit.message[:80]}')
print(f'Author: {commit.commit.author.name}')
print(f'Date: {commit.commit.author.date}')
print(f'Additions: +{commit.stats.additions}')
print(f'Deletions: -{commit.stats.deletions}')
print(f'Total changes: {commit.stats.total}')

# Large commit warning
if commit.stats.total > 500:
    print('WARNING: Large commit — careful code review needed')

Listing Files Changed in a Commit

commit.files is a list of File objects for each file changed in the commit. Properties: filename, status (added/modified/removed/renamed), additions, deletions, patch (the diff text).

from github import Github
import os

g = Github(token=os.environ['GITHUB_TOKEN'])
repo = g.get_repo('myorg/myrepo')
commit = repo.get_commit('abc12345sha')

print(f'Files changed: {len(commit.files)}')

for f in commit.files:
    print(f'  [{f.status.upper():<8}] {f.filename}')
    print(f'    +{f.additions} -{f.deletions}')
    if f.previous_filename:  # for renamed files
        print(f'    Renamed from: {f.previous_filename}')

# Find changed Python files
python_changes = [
    f for f in commit.files
    if f.filename.endswith('.py')
]
print(f'\nPython files changed: {len(python_changes)}')

Parsing the Diff (Patch) Text

The file.patch contains a unified diff. Lines starting with + are additions, - are deletions, and context lines have no prefix. Parse it to extract added lines for analysis — for example, to find new secrets, TODO comments, or anti-patterns.

def extract_added_lines(patch):
    if not patch:
        return []
    added = []
    for line in patch.split('\n'):
        if line.startswith('+') and not line.startswith('+++'): 
            added.append(line[1:])  # strip the '+' prefix
    return added

def scan_for_secrets(added_lines):
    import re
    PATTERNS = [
        (r'api[_-]?key\s*=\s*["\'][^"\']{20,}', 'potential API key'),
        (r'password\s*=\s*["\'][^"\']{6,}', 'potential hardcoded password'),
        (r'sk-[a-zA-Z0-9]{20,}', 'potential OpenAI key'),
        (r'ghp_[a-zA-Z0-9]{36}', 'potential GitHub token'),
    ]
    findings = []
    for i, line in enumerate(added_lines, 1):
        for pattern, label in PATTERNS:
            if re.search(pattern, line, re.IGNORECASE):
                findings.append({'line': i, 'type': label, 'preview': line[:80]})
    return findings

# --- demo ---
patch = (
    '@@ -1,2 +1,3 @@\n'
    ' def connect():\n'
    '+    api_key = "sk-abcdefghijklmnopqrstuvwx"\n'
    '+    password = "hunter2222"\n'
)
added = extract_added_lines(patch)
print('Added lines:', added)
for finding in scan_for_secrets(added):
    print(finding)

Aggregating Activity by Author

Group commits by author to produce developer activity reports. Count commits, lines added/deleted, and files touched per author across a time window. This is useful for team retrospectives and PR assignment decisions.

import datetime
from collections import defaultdict
from github import Github
import os

g = Github(token=os.environ['GITHUB_TOKEN'])
repo = g.get_repo('myorg/myrepo')

last_month = datetime.datetime.utcnow() - datetime.timedelta(days=30)

author_stats = defaultdict(lambda: {'commits': 0, 'additions': 0, 'deletions': 0, 'files': set()})

for commit in repo.get_commits(since=last_month):
    author = commit.commit.author.name
    author_stats[author]['commits'] += 1
    author_stats[author]['additions'] += commit.stats.additions
    author_stats[author]['deletions'] += commit.stats.deletions
    for f in commit.files:
        author_stats[author]['files'].add(f.filename)

print('Author Activity (last 30 days):')
for author, stats in sorted(author_stats.items(), key=lambda x: -x[1]['commits']):
    print(f'{author}: {stats["commits"]} commits, +{stats["additions"]} -{stats["deletions"]}, {len(stats["files"])} files')

Generating a Changelog from Commits

A changelog agent reads commits between two tags or dates, groups them by type (feat/fix/chore using conventional commit format), and generates a formatted changelog. Use an LLM to improve the description quality for each commit.

import re
import datetime
from github import Github
import os

g = Github(token=os.environ['GITHUB_TOKEN'])
repo = g.get_repo('myorg/myrepo')

def get_commits_since_tag(repo, tag_name):
    try:
        ref = repo.get_git_ref(f'tags/{tag_name}')
        tag_sha = ref.object.sha
        tag_commit = repo.get_commit(tag_sha)
        since = tag_commit.commit.author.date
        return list(repo.get_commits(since=since))[:-1]  # exclude the tag commit
    except Exception:
        return []

def parse_conventional_commit(message):
    match = re.match(r'^(feat|fix|chore|docs|test|refactor|perf|style)(\(.+?\))?: (.+)', message)
    if match:
        return match.group(1), match.group(3)
    return 'other', message.split('\n')[0][:60]

commits = get_commits_since_tag(repo, 'v1.2.0')
by_type = {'feat': [], 'fix': [], 'other': []}
for c in commits:
    msg = c.commit.message
    ctype, desc = parse_conventional_commit(msg)
    by_type.get(ctype, by_type['other']).append(desc)
print(f'Parsed {len(commits)} commits')

Identifying High-Risk Commits

High-risk commits are those that are large, touch critical files (auth, payment, security), or are made very close to a release. An agent can flag these for extra scrutiny before merging to main.

import datetime
from github import Github
import os

g = Github(token=os.environ['GITHUB_TOKEN'])
repo = g.get_repo('myorg/myrepo')

CRITICAL_PATHS = [
    'src/auth', 'src/payment', 'src/security',
    'config/', '.env', 'Dockerfile', 'requirements.txt'
]

def is_high_risk(commit):
    reasons = []

    # Large change
    if commit.stats.total > 300:
        reasons.append(f'Large change: {commit.stats.total} lines')

    # Touches critical files
    for f in commit.files:
        for path in CRITICAL_PATHS:
            if f.filename.startswith(path):
                reasons.append(f'Touches critical: {f.filename}')
                break

    return reasons

now = datetime.datetime.utcnow()
for commit in repo.get_commits(since=now - datetime.timedelta(days=1)):
    risks = is_high_risk(commit)
    if risks:
        print(f'HIGH RISK: {commit.sha[:8]} - {commit.commit.message[:50]}')
        for r in risks:
            print(f'  - {r}')

Comparing Two Commits

Use repo.compare(base, head) to get the diff between two refs (branches, tags, or commit SHAs). This returns a Comparison object with the combined stats, list of commits, and list of files changed between the two points.

from github import Github
import os

g = Github(token=os.environ['GITHUB_TOKEN'])
repo = g.get_repo('myorg/myrepo')

# Compare two branches
comparison = repo.compare('main', 'feature/new-api')

print(f'Commits ahead: {comparison.ahead_by}')
print(f'Commits behind: {comparison.behind_by}')
print(f'Status: {comparison.status}')  # ahead/behind/identical/diverged
print(f'Total commits in diff: {len(comparison.commits)}')
print(f'Files changed: {len(comparison.files)}')

# Summary of what changed
for f in comparison.files:
    print(f'  {f.status}: {f.filename} +{f.additions} -{f.deletions}')

# Compare two tags
tag_comparison = repo.compare('v1.0.0', 'v1.1.0')
print(f'Changes between v1.0.0 and v1.1.0: {len(tag_comparison.commits)} commits')

LLM-Powered Commit Summary

Feed a batch of commit messages and file names to an LLM to generate a human-readable summary of what changed in a time period. This is the core of a weekly engineering digest bot.

import anthropic
import os

client = anthropic.Anthropic(api_key=os.environ['ANTHROPIC_API_KEY'])

def summarize_commits(commits, time_period='last week'):
    commit_data = []
    for commit in commits[:30]:  # limit to avoid context overflow
        commit_data.append(
            f'- {commit.commit.message.split(chr(10))[0][:80]} '
            f'(+{commit.stats.additions}/-{commit.stats.deletions})'
        )

    commit_list = '\n'.join(commit_data)
    prompt = (
        f'Summarize what happened in this codebase {time_period} '
        f'based on these commit messages. Be concise (3-5 bullet points). '
        f'Focus on user-facing changes and significant technical work.\n\n'
        f'Commits:\n{commit_list}'
    )

    response = client.messages.create(
        model='claude-opus-4-5',
        max_tokens=400,
        messages=[{'role': 'user', 'content': prompt}]
    )
    return response.content[0].text

Detecting Sensitive Files in Commits

A security-focused commit analysis agent scans for commits that accidentally include sensitive files: private keys, .env files, configuration with credentials, or database dumps. Check commit.files for filenames matching known sensitive patterns.

import re
from github import Github
import os

g = Github(token=os.environ['GITHUB_TOKEN'])
repo = g.get_repo('myorg/myrepo')

SENSITIVE_PATTERNS = [
    r'\.env$', r'\.pem$', r'\.key$', r'id_rsa', r'credentials\.json',
    r'secret', r'password', r'\.p12$', r'keystore', r'\btoken\b'
]

def is_sensitive_filename(filename):
    for pattern in SENSITIVE_PATTERNS:
        if re.search(pattern, filename, re.IGNORECASE):
            return pattern
    return None

import datetime
last_day = datetime.datetime.utcnow() - datetime.timedelta(days=1)

for commit in repo.get_commits(since=last_day):
    for f in commit.files:
        matched = is_sensitive_filename(f.filename)
        if matched and f.status in ('added', 'modified'):
            print(f'ALERT: Sensitive file in commit {commit.sha[:8]}')
            print(f'  File: {f.filename} (matched: {matched})')
            print(f'  Author: {commit.commit.author.name}')
            print(f'  Link: {commit.html_url}')

Quick Check: commit.files vs pr.get_files()

Test your understanding of commit diff access.

Commit History Analysis Recap

Your agent can now analyze commit history comprehensively:

  • repo.get_commits(since, until, author, sha) — filter commits by time, author, or branch
  • commit.stats — additions/deletions/total across all changed files
  • commit.files — per-file details including the patch (unified diff)
  • Parse file.patch to scan added lines for secrets, TODOs, or anti-patterns
  • Group commits by author for activity reports
  • repo.compare(base, head) — diff between two refs (branches, tags, SHAs)
  • Use LLMs to generate human-readable summaries from commit message lists
  • Flag high-risk commits: large changes + critical files + close to releases

Frequently asked questions

Is the “Commit History and Diff Analysis” lesson free?

Yes — the full text of “Commit History and Diff Analysis” is free to read here on the web, and the AI Agents course includes 4 lessons in total. To practise it interactively (a built-in code editor and a 24/7 AI tutor) and unlock the rest of the AI Agents course, upgrade to CoddyKit PRO.

What will I learn in “Commit History and Diff Analysis”?

Walking commit history, extracting changed files, and summarizing diffs. You practise AI Agents with hands-on code you run directly in the browser, and a 24/7 AI tutor answers your questions as you work through the lesson.

Do I need any experience to start AI Agents?

No prior experience is required. AI Agents on CoddyKit is structured for beginners through advanced learners; this is — lesson 4 of 4, so you can start here or from the beginning and move at your own pace.

How long does the “Commit History and Diff Analysis” lesson take?

Most CoddyKit lessons take about 5–10 minutes. Each one is bite-sized and interactive, so you make steady progress and pick up exactly where you left off across the web and the app.

Can I write and run code in this AI Agents lesson?

Yes. Every AI Agents lesson includes a built-in code editor, so you write and run real code right in your browser and get instant AI feedback — no local setup required.

All lessons in this course

  1. GitHub REST API Overview
  2. Listing and Managing Issues
  3. Automated PR Review Comments
  4. Commit History and Diff Analysis
← Back to AI Agents