ABAP VERTEX Tools
ABAP Version, Code and Data Explorer, inside ABAP Development Tools.
This address is an Eclipse update site, not a page to read. There is nothing here for a browser, which is why you landed on it. Copy it and hand it to Eclipse:
Latest build 0.8.8.202610092043.
- In Eclipse, Help → Install New Software…
- Add…, paste the address above into Location, give it any name.
- Tick the category ABAP VERTEX Tools, then Next and Finish.
- Restart Eclipse when it asks.
Then Window → Show View → Other… → VERTEX, or right-click an object in the Project Explorer and pick VERTEX.
What it does
- SelecTor field labels. A Technical names / Text toggle for the selection and the result headers, remembered between sessions.
- Value Origin. Inspect FLOW, Formula and Expression analysis; the experimental BSE client uses ADT source and bundled abaplint.
- One Parts list for the object. The VERTEX Tools window keeps the parts of a class or a program - sections, methods, events, forms - at its left. The Logic diagram shows the method picked there and the Calls diagram starts its walk from the event or form picked there; Metrics and UML draw the whole object and show no list. Switching between the two neither redraws the list nor loses the choice. Nothing is picked for you: without a named part, an object opens whole.
- Versions laid out as AVE's. The parts, the versions of the chosen part under them, and the diff beside. Pin any version as the base and compare the others with it (Prev | Any); show the change in one column or two (Inline | 2 pane). AVE's switches are back: TOCs, Dups and Case/ind. The diff is a shortest edit script over whole lines with nothing on top: Eclipse's own Text Compare in Eclipse, and in VS Code VERTEX's own line differ (Myers), written from the published algorithm because Eclipse's comparer is under the EPL and VERTEX is MIT.
- Calls, Logic and UML diagrams. The flow of a program is now the Calls diagram and the branch scheme of one method the Logic diagram. Calls starts from the event or form picked in Parts, as a double-click in ACE's tree does, and draws one block per method or per class. Clicking one of those methods draws its Logic diagram inside the same canvas, joined to the block it came from, so the call and what it does are read as one picture; the open block is marked in the theme's focus colour and follows the Top-down / Left-right choice. The Logic diagram no longer drops what follows the last branch of a method, and no longer emits a node for a statement whose text is left with nothing to show — one such empty node failed the whole diagram. Blocks and decisions follow the editor's theme, and the magnifier is the explicit Lens off / Lens on toggle in the toolbar (Shift and the touchpad set its strength), rather than a rule that measured the text on screen and flickered as the diagram was zoomed.
- Help on every metric. The ? in Code Explorer explains what ACE counts, its formula, and which values are good.
- Claude models by version. The Assistant offers the newest Claude model of each family by full id, and Specify version… checks any other id with one request before keeping it.
Settings are in Window → Preferences → VERTEX Assistant. The chat needs Node.js 22 or newer and Claude Code or Codex installed and signed in: it runs on your Claude or ChatGPT subscription, which has no public API and works only through the vendor's own client.
The other half is ABAP
Not all of VERTEX waits for it. The editor, the debugger and the assistant work over ADT alone, and so do Code Explorer, Value Origin and SelecTor. In VS Code Versions and the review of a transport do too, in VS Code and in Eclipse; the hub is needed only to save a review to SAP, and without it reviews are saved to files.
| Tool | What it does | ABAP backend |
|---|---|---|
| Enhanced Code Editor (VS Code) | Hover with a data element's domain resolved, Go to (F12), Outline, Save & Activate and block-by-block Review & Activate, ABAP Unit into the Test Explorer (Ctrl+Shift+F10), ATC into Problems (Ctrl+Shift+F2), where-used (Shift+F12), SAP's keyword documentation (F1) | not needed |
| AI-driven ADT debugger (VS Code) | Breakpoints with conditions SAP evaluates and watchpoint logs, the run started in WebGUI, the stops, a short dump reported instead of silence, a verdict naming the line and the values | not needed |
| Visual Debug (VS Code; Eclipse in test) | The same session on screen: source, breakpoints, stack, every variable, tables as grids, the flow chart of a recorded run and its player | not needed |
| Value Origin | Forward Usage Analysis: how a value is computed, down from a line across calls - the static call stack, the derivation as FLOW or Formula, each as a tree or a diagram. Backward Usage Analysis: where the values of a routine go in its callers, by SAP's where-used | not needed |
| AI Assistant | Chat over the SAP system: reads, explains and changes code — the change is reviewed block by block before activation — runs the tests and ATC on an object, and drives the debugger | not needed |
| Versions / Reviewer | Version history, the diff between two versions, the review of a whole transport with approve, decline and comments | to save reviews in SAP |
| Code Explorer | Metrics (McCabe, Halstead, maintainability), UML, the Calls diagram of an object and the Logic diagram of one method; a method opened from Calls draws its Logic diagram in the same picture | not needed |
| Data Explorer (SelecTor) | Tables, views and CDS with select-options, a join built from the dictionary's foreign keys, a pivot over either | not needed |
There is one thing to pull
into the system with abapGit and activate:
VERTEX itself, its src/
folder — the hub, the resource that saves a review (ZCL_VX_ADT_RES_STORE), the
ZAVE_REVIEW table reviews are saved in.
Simple Data Explorer,
ACE,
AVE and the other projects VERTEX grew out of are
where its tools' ideas were worked out first. VERTEX does not depend on them: none of them has to
be installed, and none of their code is called. Code Explorer, Value Origin and Visual Debug read
ADT source and analyse it with abaplint in the editor; SelecTor builds its statements and reads
them through ADT's standard data preview; in VS Code the version history comes from ADT's revision feed and the review of a transport is
built from it in the window, with AVE's rules; saving it to SAP needs VERTEX's own src/,
which writes it into the ZAVE_REVIEW table AVE uses too. The AVE repository itself is not needed.
Building a code review is part of it, not only reading one. In VS Code a request with no saved
review is built from ADT when it is opened - every part it changed, the pair of versions chosen
by AVE's rule, diffed and cut into blocks - and the status line says which part is being read.
Save keeps it, in ZAVE_REVIEW or in files as vertex.review.storage
says; approving, declining and commenting then change it the way AVE does, so AVE and VERTEX read
the same review. Eclipse does the same; where it keeps reviews is set in Window → Preferences
→ VERTEX Code Review.
Pulling it is the whole step — there is nothing to register by hand. A window opened before the ABAP is there tells you the same thing.
VS Code
The same three pages run under a VS Code extension, from the same files: VERTEX ABAP on the Marketplace.
The ABAP editor and the AI-driven debugger remain VS Code only by design, while the explorer pages are shared between the two hosts. The VS Code extension also runs ABAP tests and ATC checks from chat, integrates test results with Test Explorer and Problems, provides where-used and SAP keyword documentation, and includes LLM Providers and Tools windows. The extension also hands transport reviews to the AI assistants in the editor — Copilot, Claude Code, Codex — over MCP. In VS Code, the built-in VERTEX Assistant can also use a signed-in GitHub Copilot model through the VS Code Language Model API. Its LLM Providers window has independently collapsible model lists and API-key providers for Anthropic, OpenAI, Mistral, Gemini, Groq, Cerebras, OpenRouter and NVIDIA. Each provider's model list has a Clear all / Select all control. With a local .abap file active, the chat can explain its current editor text without a SAP connection. Local ABAP programs and classes also support Outline, definition navigation, UML, metrics, Logic and Calls diagrams, and forward/backward usage analysis. Open Visual Debug on a local ABAP file shows Run with ABAP Transpiler, with radio-button selection before launch; the run stops at breakpoints and steps on from there. In VS Code Visual Debug, Formula and Expression of a chosen value are also built from a recorded run, with its values when recorded. VERTEX bundles the transpiler and runtime and generates a temporary project configuration when needed. VS Code's Debug view shows breakpoints and variables. Version history and SAP execution require a system. ABAP Unit in CI is another use case for transpiler-compatible code: a GitHub Actions Node.js job can install the project's npm dependencies, transpile the sources and run the generated tests on each push without SAP. open-abap-core provides a working example; tests requiring SAP services need substitutes or a SAP-backed job. In VS Code, Analyze Local ABAP Package on an Explorer folder shows UML and aggregate metrics for its .abap files, including subfolders. An abapGit folder tree can contain subpackages. Source lookup prefers the opened file's folder and reports duplicate object names in a selected tree. The command also runs from the VS Code Command Palette and then asks for the folder. The main VS Code VERTEX Tools window also offers Local Folder in its type list, with UML and Metrics in the same window. UML and Metrics show the current analysis stage, object count and elapsed time. Local folder analysis reuses parsed source when switching views until a file changes. UML and Metrics show the current analysis stage, object count and elapsed time. Local folder analysis reuses parsed source when switching views until a file changes. In package Metrics, Expand all / Collapse all above Objects opens or closes every object's method table. Click a method in UML to open its source in the editor. In VS Code, the source opens beside UML without switching focus. Metrics method links in VS Code behave the same way, including package tables. UML also lists interface methods in the classes that implement them. FLOW and Value Origin diagrams support graphs with more than 500 edges in both hosts. In Eclipse, Tools selects the active ADT class or program when opened from its editor. Each provider can keep multiple named API keys; the user selects the active one. For API-key providers, the VS Code chat displays token counts reported by the API across the complete answer. With vertex.ai.logPath set, the VS Code assistant writes an Anthropic-compatible JSONL log for any provider. Direct API and Copilot model calls are recorded individually; CLI subscriptions are recorded per VERTEX invocation. LLM Providers has a folder picker and a gear for log settings. The 📜 button on the VERTEX panel opens a session report with filters for date, provider, model and prompt text, plus token totals and saved answer or tool details. VERTEX registers ABAP itself; opening SAP source does not require another ABAP extension. Claude Code detection also accepts its native CLI in PATH or ~/.local/bin when the Claude Code VS Code extension is absent; restart VS Code after installation. LLM Providers checks Claude availability before offering models and reports a missing executable when settings are opened. Codex also detects a standalone native CLI or its global npm native executable through PATH without the openai.chatgpt extension. Restart VS Code after installation. Windows detection also checks %APPDATA%/npm. MCP setup offers Codex or Claude Code and copies both review and debugger registrations together. Every Welcome step includes an Info button explaining system import, connection testing, assistant choice or MCP capabilities. Open Welcome with VERTEX: Get Started in the Command Palette. Recommended optional VS Code companion: ABAP (larshp.vscode-abap). The two transport-review MCP tools use standard ADT without VERTEX ABAP classes: they read saved reviews from ZAVE_REVIEW through data preview, or build unsaved reviews from version history. They cannot save or approve reviews.
In VS Code an assistant can also debug ABAP by itself: it sets breakpoints with conditions that SAP evaluates, starts the program in WebGUI, reads what the program held where it stopped, and ends with a verdict naming the line and the values. It changes no variable and no code. The VERTEX chat has it built in; Claude Code and Codex connect to it over MCP. How it works.
Visual Debug (VS Code, pilot) puts the same debugger on screen: the source with breakpoints set by a click, the current line, the stack, every variable at once and tables in grids. It is one session with the assistant's, so what either does, the other sees. A recorded run is drawn as a top-down chart of the calls and replayed stop by stop. In a VERTEX source tab, VERTEX: Open Visual Debug reveals a right-hand panel for the current object without replacing the source, and highlights its stopped line; F5-F8 step that same session. Details.
In Eclipse, VERTEX: Open Visual Debug (it replaces Debug Monitor) shows the same page over ADT's own debug session: stepping, frames, variables, tables and breakpoints are ADT's, and Run opens the object in Eclipse's embedded SAP GUI. Eclipse navigates the source on live stops; Visual Debug does not open another class tab on every step. It is in a local test build only, not on this update site yet, and has not been checked on a live session. Details.