EDA scripting workbench
Stop rewriting the same EDA automation.
Generate and repair SKILL, Tcl, and supporting scripts with tool-version context, grounded API references, and validation evidence you can inspect.
Built for analog, mixed-signal, CAD, academic, contractor, and open-PDK workflows.
1; Batch DRC + summary extraction for a Virtuoso layout.2; Submits a rule deck to the tool runtime and writes a CSV summary.3procedure( ssRunDrcBatch( libName cellName @optional (deck "rules.drc") )4 let( ( cv markers summary )5 cv = dbOpenCellViewByType( libName cellName "layout" "maskLayout" "r" )6 unless( cv error( "cannot open %s/%s\n" libName cellName ) )78 ; dvRunDrc executes inside the vendor tool — never on our server9 markers = dvRunDrc( ?cellView cv ?deck deck ?runDir "./drc.run" )10 summary = ssSummarizeMarkers( markers )1112 ssWriteCsv( summary "drc_summary.csv" )13 dbClose( cv )14 summary15 )16)How it works
A workbench, not a chat box.
Three steps, each leaving evidence you can read.
- 01
Describe the request
Pick the vendor, tool, version, and language. State the objective in plain words. Attach the failing error output if you are repairing.
- 02
Ground against your documentation
Hybrid retrieval pulls exact API names and version notes from the docs you uploaded to the project — not a generic guess from the open web.
- 03
Inspect the script and the evidence
Read the code, the API references with their sources, and every deterministic check before you trust a single line.
Real output
Every script comes with its evidence.
The same request returns the code, the API references with sources, and the deterministic checks. Nothing is hidden behind a single score.
When an API symbol cannot be found in your documentation, it is labeled unconfirmed — not quietly presented as correct.
- grounded
dbOpenCellViewByTypeOpens a cellview for read access.
Virtuoso DB reference · §4.2
- grounded
dbCloseCloses an open cellview handle.
Virtuoso DB reference · §4.9
- unconfirmed
dvRunDrcSubmits a rule deck to the DRC runtime.
Not found in uploaded docs
- unconfirmed
ssSummarizeMarkersHelper you provide to fold the marker DB into rows.
Project helper — define before use
What it handles
The work you keep meaning to script.
DRC batch automation
Run a rule deck across cells and fold the marker database into a readable summary.
Layout scripting
Generate or parameterize layout-construction steps from a described intent.
Report extraction
Pull numbers out of tool reports into CSV or structured rows.
Netlist processing
Rename, filter, and reshape netlists without hand-editing thousands of lines.
Repetitive CAD tasks
Wrap the multi-step ritual you run by hand into one reviewable script.
Script repair
Paste a broken script and its error; get a fix with the change explained.
What we claim, exactly
Honest labels in a high-trust industry.
The product states what it checked and what it did not. It does not imply runtime verification it never performed.
Verification states
- Generated
- Produced by the model. No checks have run yet.
- Static checks passed
- Deterministic structural and safety checks ran and found no blocking issues. Not executed.
- Documentation references checked
- Referenced API symbols were compared against your uploaded documentation.
- Requires vendor-runtime testing
- Correctness depends on running inside the vendor tool. We did not and cannot run proprietary EDA runtimes.
- Customer confirmed
- You ran this in your own tool and confirmed it works.
How your data is handled
- Documents are private and scoped to a single project.
- Project content is stored only to persist your work.
- Prompts, code, and documents never enter analytics or error logs.
- Generated and uploaded code is never executed on our servers.
What it does not claim
- It does not claim to have run proprietary EDA tools.
- It does not replace a CAD or methodology engineer.
- It does not invent compatibility it cannot ground in your docs.
Pricing
One honest plan.
No usage is unlimited, and the limits are written down.
Individual
For one engineer generating and repairing EDA automation with grounded references.
Start using itTeams of more than one? Get in touch.
- 200 standard runs per month
- 20 advanced runs per month for hard repairs and escalation
- Up to 5 active projects
- Private, project-scoped document retrieval
- Deterministic validation and inspectable API references
FAQ
Direct answers.
- No. We run deterministic static checks on our server — parsing, structure, and safety scanning. We never execute proprietary EDA runtimes or your uploaded scripts. Correctness against your tool still requires running it in your tool.
- Every API symbol the script calls is compared against the documentation you uploaded to the project. Symbols found in your docs are labeled grounded with a source; symbols that are not found are labeled unconfirmed, never presented as verified.
- Project content is stored only so your projects persist between sessions. It is scoped to your account, is not used to train models, and is never sent to product analytics or error logs. You can delete a project or your whole account at any time.
- Cadence SKILL, Cadence-oriented Tcl, Synopsys-oriented Tcl, Siemens-oriented Tcl, generic Tcl, and Python utilities used around EDA workflows.
- No. It is an independent product and is not affiliated with, endorsed by, or sponsored by Cadence, Synopsys, Siemens, or any other vendor.