What's New?AI-assisted mapping is here

From n8n to Azure Logic Apps, and back

A best-effort bidirectional converter, not a black box. Every node and action runs through one shared IR — what maps cleanly converts instantly, and anything else gets a Claude-suggested fix plus a clear manual-review report, never a silent guess.

Start converting Read the docs

Runs client-side · nothing uploaded

Wworkflow-converter/poll-api.json

Search conversions…

Conversion · poll-api.jsonn8n → Azure Logic Apps · client-side, no upload

Converting92%

Steps mapped7/8

Feasibility92%

AI-assisted1

stepstatus

Parse source workflowQueued

Match node/action registryQueued

Translate expressionsQueued

Wire dependency graphQueued

Ask Claude for unmapped stepsQueued

Merge AI suggestionsQueued

Compute feasibility scoreQueued

Running: …0 / 7 steps

Built for conversions you can trust

Capability matching, structured reports, credential placeholders, and live feasibility scoring — everything that makes a best-effort conversion actually dependable.

Registry match rate83%

scheduleTriggermatched

httpRequestmatched

customNodeunmapped

Capability-based registry

Nodes are matched by what they do, not just their type name — most steps map on the first pass.

n8nLogic AppsTriggersActionsExpressions

Full coverage, both directions

Triggers, actions, expressions, and credentials — mapped n8n → Logic Apps and Logic Apps → n8n.

manual-review.json

unmapped-nodestep: Post to Slack

Medium

AI suggestion attached● reviewable

Structured review report

Every unmapped node and untranslatable expression lands in a report — severity, category, difficulty.

azureBlobStoragematched

n8n Wait → Delaymatched

HTTP actionbuilt-in

custom nodeAI fallback

Growing connector registry

Azure Blob Storage, Wait/Delay, and more — matched by capability, with new ones added the same way.

Placeholdersflagged

HTTP_HEADER_AUTH••••••

AZURE_STORAGE_KEY••••••

API_CONNECTION••••••

Credentials, never invented

Every credential reference becomes a placeholder — reconnect it on the target platform, nothing is guessed.

Feasibility score92%

+20pts

Feasibility, scored live

Start at a deterministic baseline — the score climbs as Claude resolves each unmapped step.

workflow.json

"type": "Recurrence",
"recurrence": {
  "frequency": "Minute",
  "interval": 5
}

Real target-platform JSON

Not a summary — the actual Recurrence trigger, Http action, or n8n node, ready to import.

Slack post → ApiConnectionMedium

ConfidenceHigh

Applied · markedReview

Review before you trust it

AI suggestions are marked and scored — reviewable, never applied silently.

One converter, two directions

The building blocks of a reliable conversion

From parsing the source workflow to a feasibility score — what makes an n8n ⇄ Logic Apps conversion trustworthy, not just plausible.

Bidirectional by design

One shared IR converts n8n → Logic Apps and Logic Apps → n8n through the same parsers and emitters — not two divergent one-way scripts.

Capability-based mapping

Nodes are matched by what they do, not just their name — an Azure Blob Storage node still maps to its Logic Apps connector.

Expressions, both directions

n8n's {{ $json.field }} and Logic Apps' @{body('Name')?['field']} translate directly; arbitrary code is flagged, never guessed.

Feasibility scoring

A 0–100 score and red/amber/green traffic lights on every step estimate how much manual work is actually left.

Nothing invented, everything flagged

Credentials are never fabricated. Every placeholder and untranslatable expression lands in a structured review report.

AI fills the gaps

Unmapped steps get a Claude-suggested equivalent automatically — clearly marked in the output, never silently blended in.

Ask Claude: What's the closest Logic Apps action for n8n-nodes-base.slack?

Claude handles the mappings
we haven't written yet

When a node or action has no registered mapping, Claude proposes a real target-platform equivalent automatically — clearly marked, merged into the output, and factored into the feasibility score.

Unmapped step detected

n8n-nodes-base.slack has no registered mapping

Registry + connector lookup

Checked capability-based mappings first

Asking Claude

Suggestion applied

Merged into output, marked __aiSuggestion

By the numbers:

Built-in node/action mappings

Secrets ever invented

Client-side conversion core

Convert, review, and ship
with confidence

Structured output, always

Every conversion produces valid target-platform JSON plus a structured report — nothing silently dropped, nothing silently guessed.

workflow.logicapp.json

1// the actual Logic Apps action, not a summary
2"Fetch_Status": {
3  "type": "Http",
4  "inputs": {
5    "method": "GET",
6    "uri": "...@{triggerBody()?['id']}"
7  },
8  "runAfter": { "Every_5_Minutes": ["Succeeded"] }
9}

Built on a few hard rules

Best-effort automatic conversion only works if you can trust what it did and didn't do. These are the constraints the converter never breaks.

Try the converter

Nothing invented

Credentials, secrets, and connection strings are never fabricated — every one becomes a placeholder with a flagged reconnection note.

Always reviewable

Every unmapped node, untranslatable expression, and AI suggestion lands in a structured report with a severity, category, and difficulty rating.

Capability, not naming

Nodes and actions are matched by what they actually do, so an Azure Blob Storage node still finds its Logic Apps connector.

Nothing invented

Credentials, secrets, and connection strings are never fabricated — every one becomes a placeholder with a flagged reconnection note.

Always reviewable

Every unmapped node, untranslatable expression, and AI suggestion lands in a structured report with a severity, category, and difficulty rating.

Capability, not naming

Nodes and actions are matched by what they actually do, so an Azure Blob Storage node still finds its Logic Apps connector.

Nothing invented

Credentials, secrets, and connection strings are never fabricated — every one becomes a placeholder with a flagged reconnection note.

Always reviewable

Every unmapped node, untranslatable expression, and AI suggestion lands in a structured report with a severity, category, and difficulty rating.

Capability, not naming

Nodes and actions are matched by what they actually do, so an Azure Blob Storage node still finds its Logic Apps connector.

Nothing invented

Credentials, secrets, and connection strings are never fabricated — every one becomes a placeholder with a flagged reconnection note.

Always reviewable

Every unmapped node, untranslatable expression, and AI suggestion lands in a structured report with a severity, category, and difficulty rating.

Capability, not naming

Nodes and actions are matched by what they actually do, so an Azure Blob Storage node still finds its Logic Apps connector.

AI is labelled, not blended

When Claude proposes a fix for an unmapped step, it's merged in with an __aiSuggestion marker — visible, never disguised as a native mapping.

One IR, both directions

n8n → Logic Apps and Logic Apps → n8n run through the same parsers and emitters — not two independent, drifting converters.

Scored, not guessed

A 0–100 feasibility score and red/amber/green traffic lights estimate how much manual work is actually left, per step.

AI is labelled, not blended

When Claude proposes a fix for an unmapped step, it's merged in with an __aiSuggestion marker — visible, never disguised as a native mapping.

One IR, both directions

n8n → Logic Apps and Logic Apps → n8n run through the same parsers and emitters — not two independent, drifting converters.

Scored, not guessed

A 0–100 feasibility score and red/amber/green traffic lights estimate how much manual work is actually left, per step.

AI is labelled, not blended

When Claude proposes a fix for an unmapped step, it's merged in with an __aiSuggestion marker — visible, never disguised as a native mapping.

One IR, both directions

n8n → Logic Apps and Logic Apps → n8n run through the same parsers and emitters — not two independent, drifting converters.

Scored, not guessed

A 0–100 feasibility score and red/amber/green traffic lights estimate how much manual work is actually left, per step.

AI is labelled, not blended

When Claude proposes a fix for an unmapped step, it's merged in with an __aiSuggestion marker — visible, never disguised as a native mapping.

One IR, both directions

n8n → Logic Apps and Logic Apps → n8n run through the same parsers and emitters — not two independent, drifting converters.

Scored, not guessed

A 0–100 feasibility score and red/amber/green traffic lights estimate how much manual work is actually left, per step.

1// the actual Logic Apps action, not a summary

2"Fetch_Status": {

3 "type": "Http",

4 "inputs": {

5 "method": "GET",

6 "uri": "...@{triggerBody()?['id']}"

7 },

8 "runAfter": { "Every_5_Minutes": ["Succeeded"] }

1// the actual Logic Apps action, not a summary

2"Fetch_Status": {

3 "type": "Http",

4 "inputs": {

5 "method": "GET",

6 "uri": "...@{triggerBody()?['id']}"

7 },

8 "runAfter": { "Every_5_Minutes": ["Succeeded"] }

1// the actual Logic Apps action, not a summary

2"Fetch_Status": {

3 "type": "Http",

4 "inputs": {

5 "method": "GET",

6 "uri": "...@{triggerBody()?['id']}"

7 },

8 "runAfter": { "Every_5_Minutes": ["Succeeded"] }

1// the actual Logic Apps action, not a summary

2"Fetch_Status": {

3 "type": "Http",

4 "inputs": {

5 "method": "GET",

6 "uri": "...@{triggerBody()?['id']}"

7 },

8 "runAfter": { "Every_5_Minutes": ["Succeeded"] }

From n8n to Azure Logic Apps, and back

Built for conversions you can trust

Capability-based registry

Full coverage, both directions

Structured review report

Growing connector registry

Credentials, never invented

Feasibility, scored live

Real target-platform JSON

Review before you trust it

The building blocks of a reliable conversion

Bidirectional by design

Capability-based mapping

Expressions, both directions

Feasibility scoring

Nothing invented, everything flagged

AI fills the gaps

Claude handles the mappings we haven't written yet

Unmapped step detected

Registry + connector lookup

Asking Claude

Suggestion applied

Convert, review, and ship with confidence

Built on a few hard rules

From n8n to Azure Logic Apps, and back

Built for conversions you can trust

Capability-based registry

Full coverage, both directions

Structured review report

Growing connector registry

Credentials, never invented

Feasibility, scored live

Real target-platform JSON

Review before you trust it

The building blocks of a reliable conversion

Bidirectional by design

Capability-based mapping

Expressions, both directions

Feasibility scoring

Nothing invented, everything flagged

AI fills the gaps

Claude handles the mappings we haven't written yet

Unmapped step detected

Registry + connector lookup

Asking Claude

Suggestion applied

Convert, review, and ship with confidence

Built on a few hard rules

From n8n to Azure Logic Apps, and back

Built for conversions you can trust

Capability-based registry

Full coverage, both directions

Structured review report

Growing connector registry

Credentials, never invented

Feasibility, scored live

Real target-platform JSON

Review before you trust it

The building blocks of a reliable conversion

Bidirectional by design

Capability-based mapping

Expressions, both directions

Feasibility scoring

Nothing invented, everything flagged

AI fills the gaps

Claude handles the mappings we haven't written yet

Unmapped step detected

Registry + connector lookup

Asking Claude

Suggestion applied

Convert, review, and ship with confidence

Built on a few hard rules

Claude handles the mappings
we haven't written yet

Convert, review, and ship
with confidence

Claude handles the mappings
we haven't written yet

Convert, review, and ship
with confidence

Claude handles the mappings
we haven't written yet

Convert, review, and ship
with confidence