capTable summed ALL option rows into optionShares / per-stakeholder options
regardless of status, so EXERCISED, EXPIRED and CANCELLED grants inflated
fullyDilutedShares and understated every real ownership %. A cap-table surface
must not do this.
Filter both dilution queries (the totals sum and the per-stakeholder subquery)
to OUTSTANDING options only: status NOT IN ('EXERCISED','EXPIRED','CANCELLED').
EXERCISED options already became shares (counted in `share`) — counting the
option too double-counts; EXPIRED/CANCELLED are void. DRAFT + ACTIVE still dilute.
Rebuilt goja/bundle.js. Existing goja bundle tests pass.
Claude-Session: https://claude.ai/code/session_016yg7GPhYdWCh9vpp4HEwLZ
Co-authored-by: hanzo-dev <dev@hanzo.ai>
goja/ — captable inside the unified cloud binary
bundle.js is a self-contained, ESM-free port of the captable tRPC business
logic (src/trpc/routers/{stakeholder,share-class,securities}-router), authored
in TypeScript (src/*.ts) and bundled by esbuild so it runs verbatim inside the
dop251/goja JavaScript engine embedded in
hanzoai/cloud per HIP-0106. It is the pilot
of the read-write-Base-via-goja pattern (epic #96): the same idiom sign (#98)
and dataroom (#99) will replicate.
Why a bundle?
goja runs ES2020 but not ES-module import/export and not node:
builtins — so Prisma, tRPC, next-auth and zod cannot load. Rather than rewrite
the logic in Go, the domain + validation logic lives here as TypeScript and runs
in goja; persistence is delegated to the Go host through an injected __db
bridge (per-tenant Base/SQLite). The bundle carries logic, never a DB engine.
Host contract
The Go host (hanzoai/cloud/clients/captable) injects these globals before the
first dispatch, then calls handle per request:
globalThis.__db.query(sql, args) -> row objects (SELECT)
globalThis.__db.exec(sql, args) -> { changes, lastId } (INSERT/UPDATE/DELETE)
globalThis.__newId() -> collision-resistant id
globalThis.__now() -> unix milliseconds
globalThis.handle({ route, params, orgId, session, body }) -> { status, body }
orgId is the gateway-minted, validated tenant. The host uses it BOTH to
select the per-tenant SQLite file AND passes it here as the cap-table
companyId, so every query is scoped to one tenant — physically (one file per
org) and logically (the company_id column), mirroring the tRPC
where: { companyId } scope.
Routes (the full cap-table fold)
| route(s) | HTTP (mounted by the host) |
|---|---|
company.get / company.update |
GET/PUT /v1/captable/company |
stakeholders.{list,add,update,delete} |
/v1/captable/stakeholders[/:id] |
shareClasses.{list,create,update} |
/v1/captable/share-classes[/:id] |
equityPlans.{list,create} |
/v1/captable/equity-plans |
shares.{list,add,delete,transfer} |
/v1/captable/shares[/:id], .../shares/transfer |
options.{list,add,delete} |
/v1/captable/options[/:id] |
safes.{list,create,delete} |
/v1/captable/safes[/:id] |
convertibles.{list,create,delete} |
/v1/captable/convertibles[/:id] |
rounds.{list,create,get,close} |
/v1/captable/rounds[/:id][/close] |
rounds.investments.{list,add} |
/v1/captable/rounds/:id/investments, /v1/captable/investments |
captable |
GET /v1/captable/summary (computed cap table) |
Securities issuance (shares + options), transfers (full + partial), and
rounds (a priced round issues shares to each investor and dilutes the cap
table) are all implemented over Base. The captable route computes ownership on
a fully-diluted basis (shares + options), per-class authorized-vs-issued, plus a
convertibles + rounds summary.
Out of this repo's scope by design: documents / data-room (→ dataroom #101), e-signature (→ esign #100), email. Vesting is stored (cliff/vesting years) but schedule computation is a later enhancement.
Build & test
node goja/build.mjs # regenerate bundle.js from src/*.ts (esbuild)
node goja/test/bundle.test.mjs # smoke + wiring test against a mock __db
bundle.js is committed so hanzoai/cloud can go:embed it via embed.go
(Bundle()), exactly how @hanzo/plans ships goja/bundle.js. Edit the
TypeScript, re-run the build, commit the regenerated bundle.