FAQ

What is Graph Assistant, and what can MCP do in an AtomIQ workspace?

Graph Assistant is an in-app, graph-aware AI chat over the workspace. The built-in MCP server lets external assistants read and drive the same workspace programmatically. Both need a configured workspace graph — they do not discover a host estate on their own.

How does Interactive App Viewer turn programs into apps?

Use the drag-and-drop canvas to group programs by business usage, visualize dependencies, filter the graph, and persist the workspace. An “app” here is a saved grouping for analysis — not a generated runtime application.

How do I create SQL tables from CoreIQ VSAM or IMS layouts and then load the data?

Open Create Table(s) from CoreIQ (VSAM/KSDS) or Import Legacy Schema (IMS-DB). The approved COBOL copybook is the source of truth for column types — not JDBC metadata. Preview and save the mapping, then run the batch load (UI trigger or an exported .odz pack). Segments or datasets with no created table are skipped rath

Why is the AI Query Assistant empty or not explaining my SQL error?

The assistant needs a configured AI provider and a live connection schema. It generates and refines SQL from that schema, keeps conversation context, and can explain a failed query or stored-procedure save when you use Explain on the error notification. Without a provider or a connected catalog, the chat has nothing to

How do I build a spec from AtomIQ — live workspace vs a .atq export?

If the workspace has an AtomIQ link, import operations live. If not, upload a .atq export — that path works without a live connection. Name the spec, select operations by domain and process type, optionally attach reference files, then Generate spec. Output streams and auto-saves, and Spec stores a persisted AtomIQ sna

Why does the API Integration Points wizard show a stale AtomIQ tree?

The wizard only runs when the operation has an atomiqOperationId and the parent process has a persisted AtomIQ snapshot. It never calls AtomIQ live. Refresh the import on the parent process, then reopen the wizard. Create Plan is not wired yet — use the wizard to decide and document boundaries.

What is Draft versus a numbered version, and when should I freeze before generating artifacts?

Draft is the single editable working copy. Create version freezes Draft as the next numbered snapshot (v1, v2, …) with an optional comment; numbered versions are read-only. Freeze a version before bulk AI generation so runnable artifacts stay tied to a stable snapshot. Compare any two versions, including Draft, side by

How do I read a KSDS record by key without opening the write surface?

GET /dataset-record/by-key with datasetName and keyBase64. It requires RACF Basic but is not write-gated. Missing key returns 404; no RACF permission returns 403. Insert/update/delete on /dataset-record remain write-gated.

How do I start a 3270 or 5250 session and use function keys in the browser?

Start a session from Home; it auto-connects to the admin-configured host. Use F1–F12 and Shift+F1–F12 (F13–F24), the on-screen PF bar, Tab, and Enter. Customize mappings in admin Key Mappings. This is terminal emulation — not a Screener or Next map player.

What is type-ahead, and why does the status bar show Queued?

While the host processes Enter, PF keys, or Page Up/Down, further keystrokes queue in order and drain when the keyboard unlocks. The status bar shows Queued: N with a yellow highlight. The queue clears on connection errors. This is automatic — not a macro.

How do I run or step-debug a Hub Screens Flow from Play Flow?

Enable Hub flow play, open Play Flow, pick project and method, edit the trigger JSON, then Run Flow or Start Debug (Next / Run to End / Stop). Screen-steps-only pauses on screen-invoke steps. Requires the openlegacy connector, not client-terminal-only mode.

How do I enable CDC hosting and tear down track/lock datasets?

Set ol.cdc.enabled=true plus CoreIQ URL and service token. GET /cdc/status and POST /cdc/teardown require RACF Basic. Teardown deletes .CDC.TRACK / .CDC.LOCK under the caller’s RACF identity and is idempotent; it never deletes the shared status KSDS. This is the gateway CDC hosting path, not BMC AMI IAM file-type CDC.

What does HTTP 409 on dataset-content mean?

409 means the dataset is exclusively open by another process — not a RACF deny (403) and not a missing dataset (404). Retry after the other holder closes, or inspect who has the file open on z/OS.

How do I drive IMS/MFS maps over IMS Connect?

Configure host and data-store under Global Settings → IMS Connect, enable Remote IMS Connect on the operation, and import MFS descriptors (for example sample clmm01). The path is a MID/MOD pull loop — not 3270. Same one-active-remote-session-per-process rule as CICS/MQ.

How do I record a map ScreenFlow from the Interactive Remote Player?

Open the remote session so the map grid loads descriptors at OPEN, exercise named fields and AID keys, then record to a map ScreenFlow script. You can also author Groovy or generate with AI. Execute/debug over HTTP/WebSocket. This is not a Hub terminal Screens Flow.

Is remote AS/400 / DSPF transport available yet?

No. AS400/5250 Remote Transport is planned: DSPF map ingest and a ScreenProgramHost registry like CICS/MQ and IMS Connect. It will not be 5250 terminal emulation and will not revive RPGLE-to-Groovy conversion frontends. Use OL Terminal for live 5250 today.

Does the Player run a 3270/5250 session or converted maps?

Converted maps only. Player is a map-layout grid over converted legacy code on the JVM — not a 3270/5250 emulator and not a live host session. Use OL Terminal for TN3270/TN5250, OL Screener for map-descriptor plus MQ/IMS Connect, and Hub terminal Screens Flows for live session automation. Next is offline converted UI p

How do I generate map descriptors with ol-refactor map-descriptor and goldenCheck?

Run ol-refactor map-descriptor (or Import maps in the console) so BMS+copybook or MFS sources land as JSON in the metadata DB. goldenCheck guards descriptor correctness in the golden suite. Descriptors are map definitions — not terminal-screen recordings.

Why is there only one remote session per process, and do I start the host transaction myself?

M1 allows one active remote session per process. Target transactions (for example AQ00) must be started manually on the host today. If the Player has nowhere to send maps, check Global Settings and that the host transaction is up — not just credentials.

How do I choose H2, a host database, or a modern database after migration?

OL Next runs the replatformed programs off the host. Data is a separate choice. - Sandbox (H2) — isolated play; populate from schema.sql and seed.sql. Use when you do not want a host or production store. - Legacy DB — keep data on DB2, VSAM, or IMS-DB. Connect with JDBC, or through [[product:coreiq|OL CoreIQ]] APIs ([[

What can I configure in centralized session admin?

Admin covers connectivity (sessions per IP, macros, printer, screen flip, timeouts), key mappings, homepage branding, active sessions, recorded trails (including Hub upload), and logs. Production /admin/ uses HTTP Basic (OLADMIN / OLPASSWORD). Change default credentials.

How do I record a session trail for TermIQ or Hub upload?

Administrators manage recordings under Recorded Sessions and can upload trails to Hub. Point TermIQ at the Snapshot Webhook URL for live capture, or upload the trail file. Recording is an admin/ops concern; macros are a different, browser-local feature.

How do I get from identified screens and instances to a Hub Screens Flow?

Runtime → Flow needs identified screens plus typically 1–10 completed instances. Unidentified screens can block Hub creation. AI flow analysis produces Hub artifacts from that evidence — it does not invent a flow from a single unlabeled snapshot.

What does the Runtime reporting dashboard show?

Runtime reporting summarizes captured events, identified screens, and operation instances so you can see whether traffic is matching definitions. Use it to check collection health before you try Flow → Hub.