AI Usage Policy
Effective [date of publication]
| App | TuneCraft: Automation & Site Cleanup, a Jira Cloud app built on Atlassian Forge |
|---|---|
| Marketplace listing | Not listed yet. TuneCraft is not on the Atlassian Marketplace yet, and its app key will be added here when it is. |
| Forge app ID | ari:cloud:ecosystem::app/facece90-dfd5-4792-9bff-6a5bd061a341 |
| Vendor | Plugincraft Software Inc., a company incorporated under the laws of Québec, Canada |
The short version
- TuneCraft's only AI feature is a Rovo agent, the TuneCraft helper. The AI is Atlassian's Rovo and it runs on Atlassian's platform.
- The app has no model of its own and calls no AI service outside Atlassian.
- The helper reads with the asking person's own access and never changes Jira. Its actions that prepare work only check it and say where the person confirms it.
- Plugincraft receives none of the data the helper handles, so it cannot use it to train any model.
Scope
This policy covers the AI features of TuneCraft: Automation & Site Cleanup and the AI tools used to build it. It does not cover Rovo itself, which is Atlassian's product and is governed by Atlassian's terms and policies.
AI features of the app
The TuneCraft helper works in Rovo chat on Jira sites that have Rovo. On a site without Rovo, the app's Ask Rovo buttons are hidden and nothing else in the app changes.
| Action | What it does | What it can change |
|---|---|---|
| Site findings | Reads the counts and health scores (estimated) of each tool the person may use, and Automation's biggest findings by steps saved a month. | Nothing. |
| Flow findings | Reads one flow's findings from the last scan, why each costs steps, how to fix each, and the steps in Jira in words. | Nothing. |
| Explain a listed object | Reads what a tool's list says about one field, filter, space, person or type: why it is listed and the evidence. Jira admins only. | Nothing. |
| What to clean first | Ranks findings by the steps a month each change would save (estimated), with the tools' other findings that have no step figure. | Nothing. |
| What may stop | Reads, from the last scan, the flows a flow starts, the flows that start it, and its loop, copies and shared trigger. | Nothing. |
| Prepare a watch | Checks fields, spaces or filters (at most 50) for the Watch list for 30, 60 or 90 days, and says where the person confirms. | Nothing. It starts no watch and changes nothing in Jira. |
| Prepare Optimize all | Says what the engine would save for each named flow (at most 50) and where the person makes the file. | Nothing. It makes no file and changes no flow. |
| Propose a naming convention | Checks a naming convention for filters or dashboards that the person described, and shows example names made from fixed samples. | Nothing. The person saves it on the page. |
Buttons in the app (Ask Rovo) open Rovo chat with a prompt for what the person is looking at, named by its id. The person can change the prompt before sending it, and nothing else happens until they send it. Rovo also asks the person to confirm an action that prepares work before it runs. The buttons show only on sites with Rovo.
Which model, and where data goes
The model is Atlassian's own, in Rovo. TuneCraft does not choose, host or train a model, and makes no call to an outside AI service. The helper's actions run as Forge functions on Atlassian. To answer, an action gives Rovo only what the asking person may already see: counts, ids, the findings and savings of the last scan, and the names the person's own pages already show (flow, field, filter and space names). It sends no JQL, smart value, email address or secret. The chat itself belongs to Rovo and is kept by Atlassian. The app keeps nothing from it.
TuneCraft has no MCP tools and no REST API other than the automatic scan's web trigger. That trigger takes a push of redacted flows only with the scan's key and answers with fixed replies, never with data, so no outside AI assistant can use it to read anything from TuneCraft.
Human oversight and safeguards
- The helper changes nothing in Jira. Its actions read, or prepare work and say where to confirm it. A change is made on the app's own pages, by a person, with that person's Jira rights, after the confirmation the page asks for.
- The server checks again. Each action goes through the same checks as the app's pages: who may open TuneCraft, the licence, and the clean-up tools' limit to Jira admins. A person with no access, or one the app cannot confirm, is refused.
- Who is asking comes from the platform. The site, the person and the licence come from Forge's context for the call and never from the text the model writes. A call without them is refused.
- Text people wrote is data. Flow names, field names and other texts can carry instructions aimed at a model. The helper is told to treat them as data, and every answer starts with a note that says so. No action takes JQL, a flow or a rule from the model. The one thing an action takes from the model is a naming convention, which the server checks with the same code the settings use and returns for the person to save.
- Numbers come from the engine. Savings and counts come from the app, each marked measured or estimated. The model is told not to compute any.
- Answers can be wrong. AI output may be incorrect. Check what the helper says before you act on it. The app's own pages show the evidence.
What we commit to
- The app will not send customer data to an AI service outside Atlassian. If that ever changes, we will update this policy and the Subprocessors page before the change takes effect.
- We will not use customer data to train or tune a model. Plugincraft does not receive it.
- The app's AI features explain and prepare, and a person decides. They make no decision about people on their own.
- We keep the helper's limits and checks in place, and test them with every change.
AI in building the app
Plugincraft uses AI assistants to help write and review the app's code and this website. They work on the source code and on a development Jira site with sample data. No AI assistant is given customer data. Every change is also reviewed by a review agent that did not write it, and no change is merged without that review (see Security practices).
Questions and review
Questions about this policy: write to info@plugincraft.ca. We review this policy at least once a year.