Use AI coding assistants with quality and security intact, and prove what changed
For developers, tech leads and engineering managers: a 12-chapter playbook, a real Python practice project with 10 exercises, a 20-sheet workbook, 5 fillable forms, 8 Word files and 6 quick-reference sheets, following one fintech squad through a 12-week pilot.
- Practice pack: a small real Python project with 7 tests that fail on purpose, 10 task cards with expected results, reference solutions, a patch to review, a dependency checker and a 9-commit git history to bisect
- Example AGENTS.md, CLAUDE.md, Copilot and Cursor instruction files, plus a Claude Code settings file with deny rules and a Stop hook
- A 20-sheet workbook: self-assessment and gap plan, pilot metrics with a 4-chart dashboard and traffic lights, review, decision and evidence logs, a go/no-go readiness check and dated Tool notes
- Task card, plan review, AI code review checklist and security review forms; a policy, PR policy, sign-off memo, 30 messages, 25 prompts and 9 scenario cases
The Claude Code and AI Coding Assistants Playbook · Bodhih Pro Kits · 352 by Bodhih Training, trainers to 2,000+ companies since 2008.
How do you use AI coding assistants like Claude Code safely and well?
Brief the assistant with a short instruction file and a task card, ask for a plan before code, make tests the specification, limit what it can touch with deny rules and sandboxing, review every diff tests-first, and measure against a baseline. The Claude Code and AI Coding Assistants Playbook from Bodhih Training gives you a file for each step.
Want the full free guide first? Read How to use AI coding assistants safely and well in a team.
You already feel the problem. Here is what it is costing you.
Big diffs nobody really reads
Assistants write faster than people review. Pull requests grow, review waits double, and approvals become glances.
Rules that live in prose
'Never read .env' sits in an instruction file and works most of the time. Most of the time is not a security control.
Confident and wrong
A fluent explanation of a bug, a green test that was edited to pass, a package that does not exist. Each looks fine until it ships.
No honest answer to 'is it working?'
Leaders want a number. Surveys say faster; nobody kept a baseline, and the bad weeks are missing from the slide.
Picture the next few weeks with the kit open beside you.
Imagine your next feature. Before anyone prompts, you fill a task card with five acceptance criteria, two of them failure cases, and the tests already fail for the right reason. The assistant proposes a plan; you correct one line in two minutes.
As you review the diff, you read the tests first, then the code, then the description. The deny rule has already stopped the assistant reading the environment file an issue told it to print. The checklist takes ten minutes, and you can explain every line.
At the end of the quarter, you show a dashboard with a baseline, the bad weeks and the fixes. Whether you start with the practice pack tonight or the instruction files tomorrow, every file has a filled SAMPLE so you can see what good looks like.
26 files that do the work for you
The e-book shows you how. The tools make it happen. Here is exactly what you download.
The Claude Code and AI Coding Assistants Playbook (e-book, PDF)
The guide: methods, worked examples, scripts and exercises
Start here: AI coding assistants map
The whole job in ten steps with the file for each, a 15-minute quick start, your first hour, your first week, the 30/60/90-day plan and the 'brand it in five minutes' box.
Situation router: which files to open, in which order
14 common situations, each with the files to open in order, plus the 15-minute quick start.
AI Coding Assistants Master Workbook
20 sheets: Settings, a 12-practice self-assessment feeding a gap plan, task-card tracker, pilot metrics with a dashboard of 4 charts and traffic lights, review log, decision log, evidence register, go/no-go readiness check, a six-question tool brief, studies, dated Tool notes, crosswalk, RACI, roadmap, What changed, glossary and sources.
Business case calculator (four cases, break-even)
Low, expected, high and do-nothing cases for an AI coding assistant rollout, with net value per month, one-off cost, break-even month (or 'Never') and a 12-month cumulative chart.
Task card for AI-assisted work
One-page fillable brief for any task you hand to an assistant: goal, context, constraints, acceptance criteria (with failure cases), evidence and size limit.
Plan review form (before any code changes)
One-page fillable check of an assistant's plan: files, reuse, step order, tests, risk and scope, with your corrections and the decision.
AI code review checklist (scope, correctness, security, dependencies, understanding)
One-page fillable review for every AI-assisted pull request, with severity per finding and a second-reviewer trigger for risk areas.
Security and permissions review for AI coding assistants
Two-page fillable review of accounts and data terms, permission modes, deny rules, sandboxing, secret scanning, dependency and licence controls, prompt-injection exposure and the drill result.
Readiness check (go/no-go) to expand AI coding assistants
Printable go/no-go gate: 12 criteria marked critical or not, with evidence, and the verdict rule (any critical gap = NO-GO). Mirrors the workbook's Readiness check sheet.
Leadership sign-off memo for an AI coding assistant pilot
One-page memo asking a CTO or leadership team to approve scope, tools, controls, spend limit, measurement plan and the go/no-go date, with the four-case business case summary.
AI coding assistant policy and procedure starter
A one-to-two page policy your team can adopt: approved tools and accounts, data rules, permissions, review and ownership, dependencies and licences, unattended runs, incidents, training and review cycle.
Pull-request policy and template for AI-assisted work
Team conventions (accounts, spend, session hygiene, size limit, tests, disclosure, ownership, off-limits areas, unattended runs) and a copy-ready PR template with an AI-assistance section.
Message and email library (30 messages)
30 ready-to-edit emails and chat messages for announcing a pilot, setting up repositories, tasks and pull requests, security incidents, cost, readouts, onboarding and student declarations.
Worked examples: Ethan's filled task card, plan review and review checklist
The three most-used templates filled in completely for Northgate Ledger (SAMPLE): the transfer task card, the monthly-report plan review and the review of PR-1433, so you can see what good looks like.
Scenario casebook (9 cases with model answers)
9 realistic cases from the US, Canada, the UK, Ireland, India, Australia and New Zealand, each with facts, questions and a model answer citing the kit's method and files.
AI prompt pack for coding (25 prompts)
25 tool-agnostic prompts grouped by step (understand, plan, test, review, debug, refactor, secure, measure), each with what to paste in and what to check before using the output, plus six human-review rules.
45-minute team briefing plan and knowledge-check answer key
A minute-by-minute 45-minute session to brief your team on the method, with a live demo, a practice task and the 14-question knowledge-check answer key.
Numbers that matter: AI coding assistant KPI card
The twelve numbers that show whether AI-assisted development is working: definitions, formulas, where each lives in the workbook and what to do if it moves, with Northgate Ledger's SAMPLE results.
Tool notes and rules by market (Correct at November 2026)
Dated notes for Claude Code, GitHub Copilot, Cursor and AGENTS.md (modes, instruction files, costs, plans, data terms), and the security, AI and data rules by market with official sources.
Review and security decision card (if this, do that; mistakes to avoid)
One-page decision card: if-then rules for plans, tests, PR size, risk areas, packages, injected text and unattended runs, plus the ten mistakes to avoid and a mini review flow.
Commands and settings quick reference (Claude Code, Copilot, Cursor)
One-page reference of the commands, files and settings used in the kit: Claude Code modes and slash commands, instruction-file paths for three tools, deny-rule and Stop-hook examples, and the practice-pack commands.
Practice pack: ledgerlite Python project, 10 task cards and instruction files
A small real Python project with 7 intentionally failing tests, ten task cards with expected results, reference solutions and a verifier, example AGENTS.md, CLAUDE.md, Copilot and Cursor instruction files, a Claude Code settings file with deny rules and a Stop hook, a reviewer agent, an AI-written patch to review, a dependency and licence checker and a nine-commit git history to bisect.
AI coding assistants knowledge check (LMS-ready)
14 scenario questions with four options, the correct letter and an explanation, ready to import into Bodhih LMS or any LMS.
30-day practice plan and 12-week rollout (calendar)
31 calendar events: ten practice tasks and seven habit steps over 30 days, twelve rollout milestones and two quarterly reviews.
Task board: practice plan and rollout (CSV)
29 tasks with milestone, owner, due and status, ready to import into Jira, Trello, Planner, Asana, Notion or GitHub Projects.
Chapter by chapter
Each chapter opens with a real situation, gives you a method and ends with something to do today, using the tools in your kit.
- Format
- PDF e-book + ZIP practice pack + XLSX + fillable PDF + DOCX + printable PDF + ICS + CSV
- Files
- 26 files plus a one-click ZIP
- E-book
- 87 pages
- Works with
- Python 3.9+ and pytest, git, Claude Code, GitHub Copilot, Cursor or any assistant; Excel, Google Sheets, LibreOffice, Word, Google Docs, Acrobat Reader, Outlook, Google Calendar, Jira, Trello, Asana, Bodhih LMS
- Markets covered
- US, Canada, UK, EU, Australia and New Zealand; rules dated November 2026
- Spelling
- British English with a US glossary
How agentic coding assistants work
Context, tools, permissions and the stop condition, and why rules that must never break belong in settings, not sentences.
Choose your tools and run a safe first session
Pick tools by workflow and control with a six-question brief, then run a safe first session on a branch in a read-only mode.
Get the repository ready for AI help
AGENTS.md, CLAUDE.md, Copilot and Cursor instruction files, deny rules and a test command the assistant can run.
Write tasks an assistant can finish
A five-part task card with testable acceptance criteria, including the failure cases assistants miss.
Plan, then code
Plan mode and the six plan checks, with a threshold for when planning is worth it.
Test-driven prompting
Tests as the prompt and the stop condition, and why small steps matter more than test-first order.
Review AI-written code like it matters
A size limit, tests first, review by risk and a checklist, with what the security and review research found.
Debug with an assistant, and without fooling yourself
Reproduce, narrow, root cause, fix, protect: git bisect on a real regression.
Refactor legacy code safely
Characterisation tests and small green refactoring steps, kept apart from behaviour changes.
Security: secrets, dependencies, licences and prompt injection
Secrets, insecure code, invented packages, licences and prompt injection, with enforced controls and data rules by market.
Cost, conventions and pull-request policy
What drives usage cost, the team conventions that control it, a pull-request policy and a four-case business case.
Measure impact honestly and onboard the team
A balanced scorecard against a baseline, what the controlled studies really say, the readiness check and explain-first onboarding.
Not just reading. Tools you will actually open again.
Start here
- 00-Start-Here-AI-Coding-Assistants-Map.pdf: Start here: AI coding assistants map
- Situation-Router.pdf: Situation router: which files to open, in which order
Practice pack
- Practice-Pack-Ledgerlite.zip: Practice pack: ledgerlite Python project, 10 task cards and instruction files
Workbooks
- AI-Coding-Assistants-Master-Workbook.xlsx: 20 sheets including Settings, Self-assessment, Gap plan, Task cards, Pilot metrics, Review log, Decision log, Evidence register, Readiness check, Dashboard, Tool brief, Studies, Tool notes, Crosswalk, RACI, Roadmap, What changed, Glossary, Sources
- Business-Case-Calculator.xlsx: low, expected, high and do-nothing cases with break-even month and a 12-month chart
Fillable PDF forms
- Task card for AI-assisted work
- Plan review form (before any code changes)
- AI code review checklist (scope, correctness, security, dependencies, understanding)
- Security and permissions review for AI coding assistants
- Readiness check (go/no-go) to expand AI coding assistants
Word templates
- Leadership sign-off memo for an AI coding assistant pilot
- AI coding assistant policy and procedure starter
- Pull-request policy and template for AI-assisted work
- Message and email library (30 messages)
- Worked examples: Ethan's filled task card, plan review and review checklist
- Scenario casebook (9 cases with model answers)
- AI prompt pack for coding (25 prompts)
- 45-minute team briefing plan and knowledge-check answer key
Quick-reference sheets
- Numbers that matter: AI coding assistant KPI card
- Tool notes and rules by market (Correct at November 2026)
- Review and security decision card (if this, do that; mistakes to avoid)
- Commands and settings quick reference (Claude Code, Copilot, Cursor)
Calendar, task board and LMS
- 30-day practice plan and 12-week rollout (calendar): 31 events
- Task board: practice plan and rollout (CSV): 29 tasks
- AI coding assistants knowledge check (LMS-ready): 14 scenario questions
When you have worked through it, you will be able to…
Every outcome is practised with a worked example and a tool, not just described.
- Set up a repository so any assistant starts with the right commands, rules and limits
- Write task cards and review plans so assistants build the right thing the first time more often
- Use tests as the prompt and the stop condition, and keep changes small enough to review
- Review AI-written pull requests by risk with a checklist, tests first
- Debug and refactor with an assistant without accepting unproven explanations or hidden behaviour changes
- Put enforced controls on secrets, dependencies, licences and prompt injection
- Set team conventions, spend limits and a pull-request policy
- Run a pilot with a baseline and a balanced scorecard, and decide go or no-go
Know your starting point. Prove your progress.
People who measure before and after improve faster and can show it. AssessAll, our sister assessment platform, has AI-graded assessments that pair with this kit. Take one now, work through the kit, then re-take it.
Code Review Assessment for Software Engineering Teams
Take the assessment →AssessAll assessmentEngineering Manager Judgment Assessment for Tech Leads and New Engineering Managers
Take the assessment →AssessAll assessmentData Analysis in Python Assessment with Executed Scripts for Analysts and Engineers
Take the assessment →JobularyTurn what you learn into a measured development plan (Jobulary)
Make it a growth plan →A taste of Chapter 1
The most useful idea in Anthropic's own best-practice guide is the split between instructions the model reads and instructions the tool enforces. A line in your prompt or CLAUDE.md is non-deterministic: the model usually follows it, with no guarantee. A hook, a deny rule or a managed setting is deterministic: the harness applies it every time, whatever the model decided [S32]. Every rule that must never be broken (never read the secrets file, never push to main) belongs in the second group.
The honest comparison
| Free templates and typical advice | This kit | |
|---|---|---|
| Starting point | Install a tool and start prompting | A six-question tool brief, a safe first session and instruction files the code cannot replace |
| Practice | A demo on a toy app | A real Python project with 7 failing tests, 10 task cards with expected results and reference solutions that pass |
| Rules | Advice in prose | Deny rules, hooks and CI checks for anything that must never break, with tested examples |
| Review | Skim the diff if tests pass | Size limit, tests first, review by risk and a checklist, with a planted patch to practise on |
| Evidence | Vendor percentages | Seven named studies with their settings and limits, and a 36-source register |
| Results | A survey that says 'faster' | Baseline, balanced scorecard, readiness check and a four-case business case with the low case first |
Made for you if…
- Professional developers who use or are starting to use Claude Code, GitHub Copilot, Cursor or a company-approved assistant
- Tech leads and engineering managers running a pilot or setting team conventions
- Security and platform engineers who need practical controls for coding agents
- University students on software courses whose module allows AI assistance (integrity rules first)
- Teams in the US, Canada, the UK, Ireland and the EU, Australia, New Zealand and India
Probably not for you if…
- People who do not write or read code: Claude for Non-Coders and Idea to Live App are written for you
- Anyone looking for legal advice on AI, copyright or data protection: the dated sheet points to official sources and your adviser
- Anyone expecting a productivity percentage or a guarantee: the kit shows how to measure your own team honestly
- Training for one vendor's product only: tools are named descriptively and features change monthly
Built by trainers, tested in real rooms.
Bodhih Training has designed and delivered corporate programmes since 2008. Bodhih Pro Kits package what works in those workshops into a guide plus the exact tools we use, so you can apply it the same day without a facilitator in the room.
From here to using it in three steps
- STEP 1
Tap Get the kit
Name, email and country. UPI, card or netbanking in India; international cards everywhere else. No account to create.
- STEP 2
Download instantly
Your download page opens at once, with every file and a one-click ZIP. A permanent link lands in your inbox too.
- STEP 3
Open the first tool
Start with the quick-start chapter, then the main spreadsheet or planner. Most buyers use something from the kit the same day.
Your next AI-written pull request can be one you understand
Unzip the practice pack tonight and run the tests. Whether you start with task T01 or with the instruction files for your own repository, you will see the difference in your next review.
Personal licence for the buyer's own use, including use within your own organisation and with its staff. Please don't share or resell the files. Team, company and consultant licences: solutions@bodhih.com.
What buyers say
No ratings yet. Only people who have bought this kit can rate it - every buyer gets a one-click rating link a couple of days after their purchase.
The Bodhih family, when you are ready for more
The Git and GitHub for Students and Teams Kit
Use Git and GitHub confidently for coursework, portfolios and team projects, from first co.
The Prompt Engineering Practitioner Kit
Write, test and maintain prompts that give reliable results across ChatGPT, Claude, Gemini.
The Learn to Code with AI Student Kit
Learn programming properly with AI as a tutor rather than a shortcut, as a computer scienc.
AssessAll
250+ AI-graded assessments to measure skills and earn verified credentials.
Jobulary
Individual development plans that start from evidence and stay yours.
Bodhih Training
Run this topic as a live workshop for your team, in person or online.
Pewple
Book a vetted coach for career, leadership or communication. Pay per session.
Bodhih.org pathways
Certification pathways in Applied AI, Workplace English and management.
The Claude Code and AI Coding Assistants Playbook: FAQs
Read the free guideAlready bought? Resend my linkIs this kit only for Claude Code?
No. The method works with Claude Code, GitHub Copilot, Cursor or your company's approved assistant. The practice pack includes instruction files for all three and a shared AGENTS.md, and the Tool notes sheet records each tool's version-specific facts with official links. Not affiliated with or endorsed by Anthropic, GitHub, Microsoft, Anysphere (Cursor), OpenAI or Google.
Do AI coding assistants make developers faster?
It depends on the task, the codebase and the developer. Controlled studies range from 55.8% faster on a lab task to 19% slower for experienced developers on their own mature repositories. Chapter 12 and the workbook show you how to measure your own team with a baseline instead of quoting a headline.
Is AI-generated code secure?
Not by default. Research has found vulnerable code in generated output and overconfidence in people using assistants. The kit's answer is enforced controls: deny rules, secret scanning, approved packages, a review checklist and a second reviewer for risk areas.
What should go in a CLAUDE.md or AGENTS.md file?
Only what the assistant cannot learn by reading the code: exact test and build commands, rules the code does not show, decisions made outside the repo and your definition of done. Keep it short; Anthropic's documentation recommends under 200 lines. The practice pack has working examples.
How do I review AI-written pull requests?
Set a size limit, read the tests before the code and the description last, review by risk, and never approve a line you cannot explain. The AI code review checklist form and practice task T09, a planted patch with nine problems and an answer key, teach the habit.
Do I need to know Python to use the practice pack?
You need to read basic Python and run commands in a terminal. The project is small (five modules) and everything runs with Python 3.9 or later and pytest, with no network access.
How long does it take to put in place?
The 15-minute quick start gets the practice pack running. The 30-day practice calendar takes about 15 to 40 minutes a session, and the rollout roadmap covers 12 weeks for a team pilot.
Can I use this with my team?
Yes, within your own organisation under the personal licence: brief your team with the 45-minute session plan, load the knowledge check into your LMS and adapt the policy and templates. Company and consultant licences are available from solutions@bodhih.com.
Does it work in my country?
The method is universal. The dated Tool notes and Crosswalk sheets cover the US, Canada, the UK, the EU, Australia and New Zealand with official links, correct at November 2026. Laws and vendor terms change, so check them locally with your adviser.
Is this legal or compliance advice?
No. The kit is educational. It points to official sources such as NIST, the UK NCSC, the EU AI Act and Cyber Resilience Act pages and the US Copyright Office, and its policy and memo templates are starting points for your own advisers.
How can I measure my team's code review and engineering judgement?
AssessAll, Bodhih's assessment platform, offers a Code Review Assessment for Software Engineering Teams and an Engineering Manager Judgment Assessment, so you can measure a starting point before the pilot and again after it.
How do I turn what I learn into a development plan?
Use the self-assessment and gap plan in the workbook, then turn your top three gaps into a measured individual development plan with Jobulary (jobulary.com/individual-development-plan).
How much does The Claude Code and AI Coding Assistants Playbook cost?
₹1,999 including 18% GST for buyers in India, or $49 for buyers outside India. One-time payment, no subscription, with an automatic invoice.
How do I get the files after paying?
Instantly. Your download page opens the moment payment succeeds, and a permanent download link is emailed to you. Download each file separately or everything in one ZIP.
What software do I need?
Any of the usual tools: Excel or Google Sheets for spreadsheets, Word or Google Docs for templates, Adobe Acrobat Reader, Preview, Xodo or GoodNotes for the fillable PDFs, and Google, Outlook or Apple Calendar for the calendar plan.
Can I get a refund?
Because it is a digital download, kits are not refundable once downloaded. We always refund a duplicate charge or a failed delivery, and if a file is corrupted and we cannot fix it within 3 working days. Full terms: bodhih.org/refunds.
Can my company buy this for a team?
Yes. Team and company licences, and workshops on this topic run by Bodhih Training, are available. Write to solutions@bodhih.com or call +91 99000 11601.





