HTML5 Studio

Jira + Claude Fable: AI Picks Up Tasks on Its Own

A free step-by-step guide: get your own Jira API token, connect Claude Code with three prompts — and reach the level of "send a ticket number → the AI moves it to In Progress, does the work, passes review and closes it with a comment". No payments, no sign-ups with us — you set it all up yourself.

Get the free guide Manual — preorder, price rises daily

How it will work

Level 1 — free, with this guide

You — into the Claude Code chat:

"Pick up KAN-12"

the AI reads the task in Jira itself and moves it to In Progress

Writes the code A reviewer subagent checks Findings fixed on the spot

Task in Done

with a comment in the issue: what was done, which files were touched

Level 2 — the full pipeline: the manual or a turnkey setup

Epic

"User dashboard with auth" — Fable decomposes it into waves wave1, wave2

each task of a wave lives in its own branch and its own git worktree — in parallel, no conflicts

feature/KAN-12 feature/KAN-13 feature/KAN-14

wave gate: two AI reviewers (code + architecture) → the wave is merged → only then the next wave starts

Staged deploy

automated tests → staging with health-check → production

You will set up Level 1 yourself in one evening — the guide is below. Level 2 is the manual (preorder, the price rises daily until the September 20 release) or a turnkey setup with team training.

A task by its number

You type "pick up KAN-12" — the AI reads the description and comments from Jira itself, formulates an acceptance criterion if there is none, and gets to work. No copy-pasting specs into the chat.

Statuses move themselves

The AI moves the issue To Do → In Progress → Done through the Jira transitions API. You (or your client) look at the board and see live progress — as if a whole team is working.

Review before closing

Before Done, every task is checked by a separate reviewer subagent: acceptance criterion, bugs, edge cases, code style. Findings are fixed immediately — defects never reach the board.

A comment in every issue

When closing a ticket, the AI writes a report in Jira: what was done and which files were touched. Open the issue a month later — and instantly see what changed and where.

The skill persists forever

The workflow is written to CLAUDE.md — Claude Code's memory file. A new session, tomorrow, next week — the AI remembers how to work with your board, no retraining.

Safe for your secrets

The token lives in .env on your machine, hidden from git. The workflow explicitly forbids printing it to the chat, logs or code. You can revoke the token at any moment in your Atlassian settings.

Part 1: manual prep — 4 steps

1

Register Jira and create a project

The free Jira Cloud plan (up to 10 people) at atlassian.com. A regular Kanban board is fine. Add a couple of real tasks to the project — we will test the pipeline on them. Already on Jira — skip this step.

2

Get an API token

Go to id.atlassian.com → Security → Create API token. The token is shown once — copy it right away. This is the key the AI uses to work with your board; you can revoke it there at any moment.

3

Create .env in the project root

A four-line file: JIRA_BASE_URL=https://mycompany.atlassian.net, JIRA_EMAIL=your account email, JIRA_API_TOKEN=the token from step 2, JIRA_PROJECT=the project key (e.g. KAN). And immediately add an .env line to .gitignore — secrets must never reach git.

4

Install Claude Code

npm install -g @anthropic-ai/claude-code — and sign in to your Anthropic account. Open the project folder in the terminal, run claude and pick Fable with the /model command. Done — move on to the prompts.

Part 2: three prompts — and the AI works with your Jira

Send them one by one, waiting for each to finish. Prompt #1 teaches the AI your board, #2 locks the skill into the project settings, #3 is your daily working command.

1 Prompt #1 — teach the AI your Jira

Recon without changes: the AI verifies the connection using the token from .env and studies your board — statuses, transitions, tasks.

prompt-1-recon.md

Read the .env file in the project root — it holds the Jira credentials: JIRA_BASE_URL, JIRA_EMAIL, JIRA_API_TOKEN, JIRA_PROJECT. Work with Jira through REST API v3 with Basic auth (email:token). Security rule: never print the API token to the chat, logs, code or commits.

Verify the connection and scout my project:
1. GET {JIRA_BASE_URL}/rest/api/3/myself — show the account name from the response: this confirms the connection works.
2. GET /rest/api/3/project/{JIRA_PROJECT}/statuses — what issue types and statuses my board has.
3. Search issues by JQL (project = {JIRA_PROJECT} ORDER BY created DESC) — show the latest issues: key, title, status.
4. Pick any issue and request GET /rest/api/3/issue/{key}/transitions — which status transitions are available.

Finish with a summary: account, board statuses, available transitions and the list of issues in To Do. Do not change anything in Jira yet — this is recon only.

If step 1 shows your account name — the connection works. A 401 error — check the email and token in .env.

2 Prompt #2 — lock the skill into the project settings

The AI writes the Jira workflow into CLAUDE.md — Claude Code's memory file. The skill persists in every new session.

prompt-2-workflow.md

The Jira connection works — now lock the skill in so it applies in every new session. Create (or extend) the CLAUDE.md file in the project root and write this workflow into it:

## Working with Jira
- Credentials live in .env: JIRA_BASE_URL, JIRA_EMAIL, JIRA_API_TOKEN, JIRA_PROJECT. Never print the token to the chat, logs or code. Check that .env is in .gitignore — if not, add it.
- The command "pick up {issue key}" runs the full task cycle:
  1. GET /rest/api/3/issue/{key} — read the title, description and comments. If there is no acceptance criterion, formulate one and show it to me before starting.
  2. Move the issue to In Progress (POST /rest/api/3/issue/{key}/transitions).
  3. Do the work in the project code.
  4. Hand the result to a separate reviewer subagent: acceptance criterion match, bugs, edge cases, code style. Fix the findings immediately, before closing the task.
  5. Move the issue to Done (if the board has a Review or In Testing column — use it) and leave a comment: what was done, which files were touched.
- If the task is too big for one cycle — first propose splitting it into subtasks in Jira and wait for my confirmation.
- On the command "stop" — finish the current step cleanly and record the honest task status in Jira.

Write the workflow, show the final CLAUDE.md and confirm you are ready to work.

After this prompt, check: CLAUDE.md appeared (or got extended) in the project root with a "Working with Jira" section, and .env is listed in .gitignore.

3 Prompt #3 — daily mode: a task by its number

That's it. Your daily "prompt" is now one line with a ticket number. The AI does the rest by the CLAUDE.md workflow.

prompt-3-work.md

Pick up KAN-12.

Replace KAN-12 with your issue key. Works in new sessions too: the workflow is already in CLAUDE.md.

Three levels of the pipeline — pick yours

Capability	Free guide	The manual	Turnkey setup
Jira + Claude link, tasks by number	✓ set up yourself in an evening	✓	✓ we set it up for you
Statuses, comments, AI task review	✓	✓	✓
A branch and git worktree per task	—	✓ step-by-step setup	✓ turnkey
Epics and waves: decomposition, wave gates	—	✓ templates and prompts	✓ + team training
Parallel tasks without conflicts	—	✓	✓
Two AI reviewers: code + architecture	—	✓ ready reviewer prompts	✓ + cache and auto-run
Token savings: routine on cheap models	—	✓ routing scheme	✓
Staging environment and staged deploy	—	✓ make commands and health-check	✓ turnkey
Support and warranty	—	—	✓ up to 12 months

Preorder

The Manual: "An AI Pipeline With Your Own Hands"

The entire kitchen of the pipeline that builds CHATBOSS.PRO — as a step-by-step guide. Not a course "about prompts", but a working system: scripts, workflow, reviewer prompts and the rakes we have already stepped on for you.

A separate branch and git worktree per task — agents work in parallel without stepping on each other
Epics and waves: wave1/wave2 labels, a wave gate — the next wave starts only after the previous one is merged
A commit check when closing a wave — protection against an "empty foundation" (our real incident)
Two AI reviewers: code review with P0–P3 priorities and a security gate against secret leaks + architecture review: data flows, contracts, overengineering
Ready-made reviewer system prompts and task templates — the same ones running in our production
One command for everything: make task, make epic-queue, make review, make staged-deploy
Staged deploy: build → automated tests → staging with health-check → production in maintenance mode
Token savings: model routing and review cache — re-checking the same code is free
A pipeline dashboard: epics, waves, task statuses and the commit feed on one screen

Pay and get access

TON · USDT · Bitcoin · instant access

Payment in crypto (TON, USDT, Bitcoin) to the studio wallet — fast and from anywhere in the world. Once a TON payment is confirmed, access to the manual opens automatically. The price rises every day until the September 20 release (final — 29,900 ₽): pay now and you lock in today's price and get all the chapters first.

Full manual page: comparison, deliverables, FAQ →

For LinkedIn readers

15% off a turnkey pipeline setup

Don't feel like doing it yourself, even with the manual? We set up the full pipeline turnkey: git worktrees, waves and wave gates, two AI reviewers, staging and production environments, team training. Came from LinkedIn — mention the promo code and get 15% off the Launch and Partner packages.

Promo code LINKEDIN15

Get the discount See pipeline packages

This pipeline is not theory

The full version of the pipeline builds CHATBOSS.PRO — a platform for creating Telegram bots in a visual block editor. Every feature travels the route: Fable decomposition → waves of tasks → two AI reviewers → staging → production.

500+

users have already built their bots

Visit CHATBOSS.PRO → Ask the AI consultant about the pipeline

💬 The AI consultant itself runs on the CHATBOSS.PRO platform — ask it about the pipeline, timelines and pricing right in Telegram.

Frequently Asked Questions

Is the guide really free?

Yes. All four steps and three prompts are right here, no sign-up on our site and no hidden conditions. It is a working lite version of the same approach we use ourselves and roll out for clients in full.

Is it safe to give the AI a Jira key?

You put the token into a local .env file yourself and hide it from git via .gitignore — the AI only reads it from there. The workflow explicitly forbids printing the token to the chat, logs or code. And you can revoke the token at any moment at id.atlassian.com — the pipeline will simply ask for a new one.

What do I need to make it work?

Jira Cloud (the free plan for up to 10 people is enough), Claude Code, and an Anthropic subscription with access to the Fable model. Everything is laid out step by step above — the setup takes one evening, after that "pick up a task" takes seconds.

What is in the manual and when does it ship?

The manual is a step-by-step guide to the full version of the pipeline: a branch and git worktree per task, epics and waves with wave gates, two AI reviewers with ready prompts, staged deploy, token savings. Release — September 20; chapters ship as they are ready, preorder buyers get them first. The preorder price rises every day until release (final — 29,900 ₽): order now and you lock in today's price.

How do I preorder and pay?

Click "Pay and get access" — a crypto payment page opens (TON, USDT or Bitcoin). Payment goes to the studio wallet; a TON payment is confirmed automatically and access to the manual opens immediately. No crypto? TON, USDT and BTC can be bought in a couple of minutes on any exchange, in the Telegram wallet or an exchanger. Payment questions — @paul_cop on Telegram.

How is the manual different from the turnkey setup?

The manual — you do it yourself following our guide. The turnkey setup — we come to you: deploy the pipeline for your stack, train the team and give a warranty of up to 12 months. Packages from $2,000, full description on the pipeline page.

How do I get the 15% discount?

Write to us and mention the promo code LINKEDIN15. The discount applies to the turnkey pipeline setup — the Launch and Partner packages. The final quote is calculated after a free audit of your project.

Will this work for my project?

The Jira + Claude combo is universal: web, bots, mobile apps, legacy code. If you are unsure about the full pipeline — write to us, we will run a free audit and tell you honestly whether it will pay off for you.

Ready for the full pipeline?

Packages from $2,000: Jira, git worktrees, two AI reviewers, environments and deploy — turnkey. 15% off with code LINKEDIN15.

Packages & pricing

Set up the pipeline: manual or turnkey with 15% off

Tell us about your project — we'll estimate the timeline and suggest a solution

Contact us