tackle2-addon-ai-migrator

A Tackle2 addon that uses Goose to perform AI-assisted code migrations. The addon clones an application's source repo, fetches migration rules, runs a Goose recipe, and uploads the report back to the hub.

The following is a guide to setting up and testing the addon locally.

Prerequisites

• Auth disabled on the hub (for now)
• oc port-forward to the hub service
• The Addon CR applied to the cluster

Setup

1. Port-forward the hub

The hub API isn't directly accessible through the OpenShift route. Port-forward to bypass it:

oc port-forward svc/tackle-hub 8080:8080 -n konveyor-tackle

2. Apply the Addon CR (if not applied)

oc apply -f ai-migrator-addon-cr.yaml -n konveyor-tackle

Verify:

oc get addon ai-migrator -n konveyor-tackle -o yaml

Usage

1. Find an application ID

curl -sS "http://localhost:8080/applications?limit=50" \
  -H "Accept: application/json" \
  | python3 -c "
import sys, json
for app in json.load(sys.stdin):
    print(f'  id={app[\"id\"]}  name={app[\"name\"]}')
"

2. Create a task

Task instances are created via the hub REST API (not Kubernetes CRs). Leave state blank (defaults to Created) so the task manager won't try to launch the container -- for local debugging we run the addon directly.

curl -sS -X POST http://localhost:8080/tasks \
  -H "Content-Type: application/json" \
  -H "Accept: application/json" \
  -d '{
    "name": "ai-migrator-test",
    "addon": "ai-migrator",
    "application": {"id": 231},
    "data": {
      "profile": {"id": 1},
      "sourceTech": "PatternFly 5",
      "targetTech": "PatternFly 6"
    }
  }'

Save the id from the response (e.g. 1145).

Task data fields

Field	Description
profile	Analysis profile reference (by ID). Populates rules from the profile's repository/files.
sourceTech	Source technology (e.g. "PatternFly 5")
targetTech	Target technology (e.g. "PatternFly 6")
rules	Optional. Rules directly (repository, rulesets, files). If both profile and rules are set, the profile populates first, then direct rules override.

3. Run the addon locally

Update .env with the task ID from step 2:

TASK=1145
SHARED_PATH=<path_to>/tackle2-addon-ai-migrator/shared
HUB_BASE_URL=http://localhost:8080
GOOSE_BIN=<path_to>/tackle2-addon-ai-migrator/bin/fake-goose
RECIPE_PATH=<path_to>/tackle2-addon-ai-migrator/recipes/goose/recipes/migration.yaml

Note

Set GOOSE_BIN to bin/fake-goose for fast iteration. Build it with make fake-goose.

Then either hit F5 in VS Code (uses .vscode/launch.json) or:

source .env && go run ./cmd

4. Inspect results

Task status:

curl -sS http://localhost:8080/tasks/1145 \
  -H "Accept: application/json" | python3 -m json.tool

Task attachments (report file):

curl -sS http://localhost:8080/tasks/1145 \
  -H "Accept: application/json" \
  | python3 -c "
import sys, json
task = json.load(sys.stdin)
for a in (task.get('attached') or []):
    print(f'  id={a[\"id\"]}  name={a.get(\"name\",\"\")}')
"

Download the report:

curl -sS http://localhost:8080/files/<fileId> -o report.html

Query facts:

curl -sS http://localhost:8080/applications/231/facts/ai-migrator: \
  -H "Accept: application/json" \
  | python3 -c "
import sys, json
facts = json.load(sys.stdin)
for k, v in facts.items():
    if isinstance(v, list):
        print(f'{k}: {bytes(v).decode()}')
    else:
        print(f'{k}: {v}')
"

Note

Fact values come back as byte arrays due to a hub serialization issue. The python snippet above decodes them. The primary report access path is the task attachment (addon.Attach), not facts.

Goose output log:

The hub SDK's command.New() automatically captures goose stdout/stderr and uploads it as a task attachment. Find the goose output file in the attachments list (it will be named goose.output or fake-goose.output), then download it by file ID:

curl -sS http://localhost:8080/files/<fileId>

Building

make cmd             # build the addon binary
make fake-goose      # build the fake goose binary for testing
make image-podman    # build the container image

Addon Flow

1. addon.DataWith(&data)         -- parse task JSON into Data struct
2. applyProfile(data)            -- if profile ID set, fetch it and populate Rules
3. FetchRepository(application)  -- clone the app's git repo
4. Rules.Build()                 -- fetch rules from hub (files, rulesets, git repos)
5. Goose.Run()                   -- goose run --recipe ... --params ...
6. uploadReport()                -- addon.File.Post() + addon.Attach() + store facts