Sanad

Engineering Intelligence Workbench

Sanad is an Engineering Intelligence Platform that generates, validates and governs the digital thread of a product, keeping requirements, models, code, tests and certification evidence continuously aligned. It is a build in progress: a free beta, well short of complete. Nothing is called complete until its signed verification record exists.

Sanad — the Engineering Intelligence Workbench for regulated systems

Sanad turns what an engineering team already has — requirements documents, design notes, code and tests, in Git — into a verified, traceable understanding of the system, and generates the artifacts a regulated programme needs from it: requirements, SysML v2 models, diagrams, reports and certification evidence. It runs inside VS Code. Every fact it shows carries its chain back to a file at a commit. It never writes anything an engineer did not confirm.

  Git artifacts (the only truth)          Sanad (derived, never stored)              What you get
  ──────────────────────────────          ────────────────────────────               ────────────
  requirements · design notes    ───▶     Engineering Graph: every requirement,  ───▶  traceability · test coverage · impact
  SysML v2 diagrams · code · tests        part, interface, test and finding,          diagrams · reports · CI gate · baselines
                                          joined by id, rebuilt at any commit          certification evidence, byte-identical

Built for every engineer involved in developing a regulated product under DO-178C, ARP4754A, ISO 26262 or similar guidance, and for any enterprise product that wants the same discipline: systems engineers, software leads, verification engineers, reviewers and DERs. Plain files, plain Git, no server, no model database.

Start wherever your product is

New product	Work in progress	Existing product
Describe the system; Sanad drafts requirements and SysML with you, confirms each step, and the evidence trail starts on day one	Point Sanad at the repository; it reads what exists, shows the gaps in trace and coverage, and you fill them in place	Reverse-engineer: code and documents become facts, behaviours and questions; you confirm the understanding; requirements, models and reports are generated from it

What Sanad does today — build in progress

Every capability below is a build in progress, or planned. Testing by hand is under way, and nothing is called complete until its signed verification record exists.

Capability	In one line	Status
Requirements	Author, review and explore requirements as Markdown with patterns (EARS), quality findings and a per-requirement commit hash	build in progress
Traceability	Requirement → design → code → test links from `@implements` / `@verifies` markers; trace and impact views; traceability report and CI gate	build in progress
Test coverage	Coverage items per requirement, counted from the tests that actually exist; not covered is said, never guessed	build in progress
Analysis and findings	Rule packs the organisation configures; findings with evidence; HTML export that regenerates byte-identically at a commit	build in progress
Baselines and drift	Freeze a baseline, compare later, see what moved	build in progress
Workspaces	Persistent engineering sessions on a canvas: requirements, notes and diagrams as cards, bound to a branch	build in progress
Design canvas and SysML v2	Describe a system by voice or text; Sanad proposes notes and a SysML v2 diagram; you confirm; the card draws it; six diagram kinds; the model feeds the graph	build in progress
AI assistant	Proposes, asks and grounds on your workspace; writes only what you confirm; your own LLM endpoint, key never in Git	build in progress
Command line	The same analysis, gate and fact pack from a terminal and CI	build in progress
Test-case drafting	Coverage items → conditions → pytest drafts, engineer closes each step	build in progress
Discovery mode	Document → facts → behaviours → questions → engineer confirmation → discovery graph → SysML and reports as projections	planned

Analysis — thirteen engines

Validation · Quality · Traceability · Structure · Verification · Implementation · Safety · Security · Architecture · Consistency · Interface · Conformance · Impact

Findings appear as squiggles and in the Problems panel, on the exact words they name. A finding derived from prose renders as a candidate for human judgement, never as fact.

What makes it different

Sanad	Typical MBSE or ALM tool
Git files are the only truth; everything else is derived and can be rebuilt at any commit	a model database beside the code
AI proposes, a person confirms, nothing is written silently	AI output lands in the model unreviewed
Guidance vocabulary only: DO-178C, ARP4754A, SysML v2 words in the product	tool-specific terms
The organisation configures every rule; Sanad recommends defaults	rules fixed by the vendor
An independent verification record per feature, kept as the build progresses	verification claimed, not evidenced

How to start (10 minutes)

Install the extension → Sanad: Set Up Sanad in This Folder → open a requirement → see its trace, coverage and findings. Add a workspace, describe your system, confirm the proposal, watch the card draw.

Read the User Guide — 15 minutes, and section 4 on roles is the one that makes everything else make sense.

Support

Need	Where	Target
Bug	Report it	72 hours
Feature or complex issue	Request it	about one week
Question	User Guide	—
Security	SECURITY.md — never a public issue	72 hours to acknowledge

Full policy: SUPPORT.md.

Licence

Sanad is free to use during its beta period. Sanad is provided free of charge during its beta period, which covers every release numbered below 1.0. Individuals and organisations may install, use and evaluate beta releases without a licence key. Each beta release is valid for 45 days from its release date; when a release expires, install the current release to continue. The application will indicate when a licence key becomes required. Licensed use begins with version 1.0.

Proprietary — see LICENSE.

Ejadah AI Labs · test baseline 0.6.5 · build in progress · engineering standards and principles at ejadah-foundation.

Sanad — Engineering Intelligence Workbench

Ejadah AI LABS