2026-05-05 23:53:19 +00:00
|
|
|
import { expect, setDefaultTimeout, test } from 'bun:test'
|
2026-05-03 18:03:40 +00:00
|
|
|
|
fix: validate parsed JSON at untrusted boundaries with Valibot (#855)
* fix: validate parsed JSON at untrusted boundaries with Valibot
Clipboard HTML, library revisions from shared storage, MCP and automation
WebSocket messages, the MCP discovery file, sidecar output and AI/MCP tool
arguments were JSON.parse'd and cast to their expected types, so a
malformed payload reached the document or crashed paste. They now go
through v.pipe(v.string(), v.parseJson(), Schema), which reports bad JSON
and a wrong shape as the same validation failure.
The path_set tool rejects an invalid VectorNetwork and shares its parser
with create_vector. The CLI library catalog validates its files and runs
revisions through the same size, identity and content-hash checks as the
app; reading image bytes as index-keyed records also stops them coming
back empty. Hand-rolled typeof readers for plugin data, document metadata,
caches and preferences become schemas with their behaviour preserved, and
readCacheJSON takes a schema for its payload.
open-pencil/no-unvalidated-json-parse rejects type assertions on
JSON.parse results other than `as unknown` in src and packages/*/src.
* refactor: validate parsed JSON in tests and tooling
Extend open-pencil/no-unvalidated-json-parse beyond source: tests, helpers and repo tooling now parse JSON through Valibot schemas instead of asserting a type. The shared fixture reader returns a validated object; its old array annotation never matched the fixtures.
* fix: validate clipboard geometry bytes, library images and model catalogs
Clipboard geometry blobs and library image bytes must be bytes at contiguous indexes, so out-of-range or gapped values are rejected instead of silently becoming different geometry or images; serialized library nodes must carry source metadata. The models.dev and OpenRouter responses are validated like their cached copies, and activate-tab rejects a CDP frame it cannot read instead of hanging.
* refactor: extend the JSON validation lint to .json() results
no-unvalidated-json-parse now also rejects type assertions on Response, Bun.file and shell .json() results, the same unchecked parse in another form. MCP server tests read /health through a validated readHealth helper and discovery files through parseDiscoveryInfo; the remaining tooling reads its JSON through schemas.
* test: validate the RPC request body in the CLI app export test
* test: validate CLI JSON output in the tool and app command tests
* test: compare the malformed models.dev fallback with the curated list
2026-10-04 17:01:50 +00:00
|
|
|
import * as v from 'valibot'
|
|
|
|
|
|
2026-05-05 23:05:02 +00:00
|
|
|
import { cliSourcePath, repoPath } from '#tests/helpers/paths'
|
|
|
|
|
import { heavy } from '#tests/helpers/test-utils'
|
2026-03-04 12:02:06 +00:00
|
|
|
|
|
|
|
|
setDefaultTimeout(30_000)
|
Unify tool definitions: define once, adapt for AI/CLI/MCP
Move tool logic to @open-pencil/core/tools/schema.ts as framework-agnostic
ToolDef objects. AI adapter generates valibot schemas + Vercel AI tool()
wrappers automatically.
26 tools (was 10): create_shape, render (JSX), set_fill, set_stroke,
set_effects, set_layout, set_constraints, update_node, delete, clone,
rename, reparent, group/ungroup, find_nodes, get_node, get_page_tree,
get_selection, select, list_pages, switch_page, list_variables,
list_collections, create_component, create_instance, eval.
src/ai/tools.ts: 269→28 lines (adapter only).
Tests for all 3 interfaces:
- 26 core tool tests (FigmaAPI directly)
- 11 AI adapter tests (valibot + Vercel AI SDK tool())
- 12 CLI integration tests (eval command on .fig fixture)
323 total, all passing.
2026-03-01 11:59:19 +00:00
|
|
|
|
2026-05-05 23:05:02 +00:00
|
|
|
const CLI = cliSourcePath('index.ts')
|
|
|
|
|
const FIXTURE = repoPath('tests/fixtures/gold-preview.fig')
|
Unify tool definitions: define once, adapt for AI/CLI/MCP
Move tool logic to @open-pencil/core/tools/schema.ts as framework-agnostic
ToolDef objects. AI adapter generates valibot schemas + Vercel AI tool()
wrappers automatically.
26 tools (was 10): create_shape, render (JSX), set_fill, set_stroke,
set_effects, set_layout, set_constraints, update_node, delete, clone,
rename, reparent, group/ungroup, find_nodes, get_node, get_page_tree,
get_selection, select, list_pages, switch_page, list_variables,
list_collections, create_component, create_instance, eval.
src/ai/tools.ts: 269→28 lines (adapter only).
Tests for all 3 interfaces:
- 26 core tool tests (FigmaAPI directly)
- 11 AI adapter tests (valibot + Vercel AI SDK tool())
- 12 CLI integration tests (eval command on .fig fixture)
323 total, all passing.
2026-03-01 11:59:19 +00:00
|
|
|
|
|
|
|
|
async function evalCode(
|
|
|
|
|
code: string
|
|
|
|
|
): Promise<{ stdout: string; stderr: string; exitCode: number }> {
|
2026-09-17 07:18:16 +00:00
|
|
|
const proc = Bun.spawn([process.execPath, CLI, 'eval', FIXTURE, '--code', code, '--json'], {
|
Unify tool definitions: define once, adapt for AI/CLI/MCP
Move tool logic to @open-pencil/core/tools/schema.ts as framework-agnostic
ToolDef objects. AI adapter generates valibot schemas + Vercel AI tool()
wrappers automatically.
26 tools (was 10): create_shape, render (JSX), set_fill, set_stroke,
set_effects, set_layout, set_constraints, update_node, delete, clone,
rename, reparent, group/ungroup, find_nodes, get_node, get_page_tree,
get_selection, select, list_pages, switch_page, list_variables,
list_collections, create_component, create_instance, eval.
src/ai/tools.ts: 269→28 lines (adapter only).
Tests for all 3 interfaces:
- 26 core tool tests (FigmaAPI directly)
- 11 AI adapter tests (valibot + Vercel AI SDK tool())
- 12 CLI integration tests (eval command on .fig fixture)
323 total, all passing.
2026-03-01 11:59:19 +00:00
|
|
|
stdout: 'pipe',
|
|
|
|
|
stderr: 'pipe'
|
|
|
|
|
})
|
|
|
|
|
const [stdout, stderr] = await Promise.all([
|
|
|
|
|
new Response(proc.stdout).text(),
|
|
|
|
|
new Response(proc.stderr).text()
|
|
|
|
|
])
|
|
|
|
|
const exitCode = await proc.exited
|
|
|
|
|
return { stdout: stdout.trim(), stderr: stderr.trim(), exitCode }
|
|
|
|
|
}
|
|
|
|
|
|
fix: validate parsed JSON at untrusted boundaries with Valibot (#855)
* fix: validate parsed JSON at untrusted boundaries with Valibot
Clipboard HTML, library revisions from shared storage, MCP and automation
WebSocket messages, the MCP discovery file, sidecar output and AI/MCP tool
arguments were JSON.parse'd and cast to their expected types, so a
malformed payload reached the document or crashed paste. They now go
through v.pipe(v.string(), v.parseJson(), Schema), which reports bad JSON
and a wrong shape as the same validation failure.
The path_set tool rejects an invalid VectorNetwork and shares its parser
with create_vector. The CLI library catalog validates its files and runs
revisions through the same size, identity and content-hash checks as the
app; reading image bytes as index-keyed records also stops them coming
back empty. Hand-rolled typeof readers for plugin data, document metadata,
caches and preferences become schemas with their behaviour preserved, and
readCacheJSON takes a schema for its payload.
open-pencil/no-unvalidated-json-parse rejects type assertions on
JSON.parse results other than `as unknown` in src and packages/*/src.
* refactor: validate parsed JSON in tests and tooling
Extend open-pencil/no-unvalidated-json-parse beyond source: tests, helpers and repo tooling now parse JSON through Valibot schemas instead of asserting a type. The shared fixture reader returns a validated object; its old array annotation never matched the fixtures.
* fix: validate clipboard geometry bytes, library images and model catalogs
Clipboard geometry blobs and library image bytes must be bytes at contiguous indexes, so out-of-range or gapped values are rejected instead of silently becoming different geometry or images; serialized library nodes must carry source metadata. The models.dev and OpenRouter responses are validated like their cached copies, and activate-tab rejects a CDP frame it cannot read instead of hanging.
* refactor: extend the JSON validation lint to .json() results
no-unvalidated-json-parse now also rejects type assertions on Response, Bun.file and shell .json() results, the same unchecked parse in another form. MCP server tests read /health through a validated readHealth helper and discovery files through parseDiscoveryInfo; the remaining tooling reads its JSON through schemas.
* test: validate the RPC request body in the CLI app export test
* test: validate CLI JSON output in the tool and app command tests
* test: compare the malformed models.dev fallback with the curated list
2026-10-04 17:01:50 +00:00
|
|
|
const RecordJSON = v.pipe(v.string(), v.parseJson(), v.record(v.string(), v.unknown()))
|
|
|
|
|
|
test: typecheck the test suites and fix what that found (#896)
* build: typecheck the test suites
Tests were in no TypeScript program: no tsconfig included tests/** or
packages/*/tests/**, and bun strips types without checking them, so a
fixture could drop a required field and keep passing until something
read it.
@types/bun moves to the root because it was installed per package only,
and #cli-tests/* joins the paths the root config already carries.
* test: fix the type errors the test suites were hiding
Typechecking the tests turned up 1123 errors. Most were ordinary
strictness, but some were real: `NodeChange` bound to Figma's plugin
typings rather than the Kiwi codec in thirteen .fig tests,
materializeInstance was called with six arguments against five so the
blobs and source children were dropped, CanvasKit pixels were written
to a plain object that never reached WASM, and assertions were made
through accessors that do not exist, so they asserted nothing.
Fixtures that had quietly lost a required field now carry it, nullable
results are narrowed through the existing expectDefined helper rather
than assumed, and stand-ins for CanvasKit and the editor go through one
named helper instead of an unexplained cast at each site.
No test was deleted, skipped, or weakened, and no `any`, non-null
assertion, or ts-expect-error was introduced.
* docs: record what typechecking the tests established
Pins the app program's global types with an assertion rather than a
note, since an unpinned types list lets any root @types package decide
which platform src/** is judged against.
The two environment faults that look like code regressions — Vite's
dependency pre-bundle outliving a package rebuild, and heavy .fig
suites failing under load — go to the development docs, where an
explanation belongs.
* fix: align @types/bun and keep node types resolvable when extended
The root manifest declared a newer @types/bun than every package, which
check:monorepo rejects, and pinning the app program's types left them
unresolvable from a config that extends this one out of tree.
* fix: fail the test typecheck when the compiler itself fails
The gate matched diagnostics by substring, so a compiler or config
failure that named no test file printed a pass while having checked
nothing. Diagnostics are now split by whether they name a file: an
unscoped one is the run failing and stops the gate, a test file's is a
finding, and a source file's stays out by design.
Also drops the parameter planComponentConstruction never read, and
makes the inner-shadow verification script exit non-zero when it
renders no image instead of logging and succeeding.
* chore: merge master into tests-typecheck
2026-10-05 12:42:38 +00:00
|
|
|
const SolidFills = v.array(v.object({ color: v.object({ r: v.number() }) }))
|
|
|
|
|
|
fix: validate parsed JSON at untrusted boundaries with Valibot (#855)
* fix: validate parsed JSON at untrusted boundaries with Valibot
Clipboard HTML, library revisions from shared storage, MCP and automation
WebSocket messages, the MCP discovery file, sidecar output and AI/MCP tool
arguments were JSON.parse'd and cast to their expected types, so a
malformed payload reached the document or crashed paste. They now go
through v.pipe(v.string(), v.parseJson(), Schema), which reports bad JSON
and a wrong shape as the same validation failure.
The path_set tool rejects an invalid VectorNetwork and shares its parser
with create_vector. The CLI library catalog validates its files and runs
revisions through the same size, identity and content-hash checks as the
app; reading image bytes as index-keyed records also stops them coming
back empty. Hand-rolled typeof readers for plugin data, document metadata,
caches and preferences become schemas with their behaviour preserved, and
readCacheJSON takes a schema for its payload.
open-pencil/no-unvalidated-json-parse rejects type assertions on
JSON.parse results other than `as unknown` in src and packages/*/src.
* refactor: validate parsed JSON in tests and tooling
Extend open-pencil/no-unvalidated-json-parse beyond source: tests, helpers and repo tooling now parse JSON through Valibot schemas instead of asserting a type. The shared fixture reader returns a validated object; its old array annotation never matched the fixtures.
* fix: validate clipboard geometry bytes, library images and model catalogs
Clipboard geometry blobs and library image bytes must be bytes at contiguous indexes, so out-of-range or gapped values are rejected instead of silently becoming different geometry or images; serialized library nodes must carry source metadata. The models.dev and OpenRouter responses are validated like their cached copies, and activate-tab rejects a CDP frame it cannot read instead of hanging.
* refactor: extend the JSON validation lint to .json() results
no-unvalidated-json-parse now also rejects type assertions on Response, Bun.file and shell .json() results, the same unchecked parse in another form. MCP server tests read /health through a validated readHealth helper and discovery files through parseDiscoveryInfo; the remaining tooling reads its JSON through schemas.
* test: validate the RPC request body in the CLI app export test
* test: validate CLI JSON output in the tool and app command tests
* test: compare the malformed models.dev fallback with the curated list
2026-10-04 17:01:50 +00:00
|
|
|
function parseRecord(stdout: string): Record<string, unknown> {
|
|
|
|
|
return v.parse(RecordJSON, stdout)
|
Unify tool definitions: define once, adapt for AI/CLI/MCP
Move tool logic to @open-pencil/core/tools/schema.ts as framework-agnostic
ToolDef objects. AI adapter generates valibot schemas + Vercel AI tool()
wrappers automatically.
26 tools (was 10): create_shape, render (JSX), set_fill, set_stroke,
set_effects, set_layout, set_constraints, update_node, delete, clone,
rename, reparent, group/ungroup, find_nodes, get_node, get_page_tree,
get_selection, select, list_pages, switch_page, list_variables,
list_collections, create_component, create_instance, eval.
src/ai/tools.ts: 269→28 lines (adapter only).
Tests for all 3 interfaces:
- 26 core tool tests (FigmaAPI directly)
- 11 AI adapter tests (valibot + Vercel AI SDK tool())
- 12 CLI integration tests (eval command on .fig fixture)
323 total, all passing.
2026-03-01 11:59:19 +00:00
|
|
|
}
|
|
|
|
|
|
2026-03-05 05:56:35 +00:00
|
|
|
heavy('CLI tool operations via eval', () => {
|
Unify tool definitions: define once, adapt for AI/CLI/MCP
Move tool logic to @open-pencil/core/tools/schema.ts as framework-agnostic
ToolDef objects. AI adapter generates valibot schemas + Vercel AI tool()
wrappers automatically.
26 tools (was 10): create_shape, render (JSX), set_fill, set_stroke,
set_effects, set_layout, set_constraints, update_node, delete, clone,
rename, reparent, group/ungroup, find_nodes, get_node, get_page_tree,
get_selection, select, list_pages, switch_page, list_variables,
list_collections, create_component, create_instance, eval.
src/ai/tools.ts: 269→28 lines (adapter only).
Tests for all 3 interfaces:
- 26 core tool tests (FigmaAPI directly)
- 11 AI adapter tests (valibot + Vercel AI SDK tool())
- 12 CLI integration tests (eval command on .fig fixture)
323 total, all passing.
2026-03-01 11:59:19 +00:00
|
|
|
test('create and read back a node', async () => {
|
|
|
|
|
const { stdout, exitCode } = await evalCode(`
|
|
|
|
|
const r = figma.createRectangle()
|
|
|
|
|
r.name = 'TestRect'
|
|
|
|
|
r.x = 100
|
|
|
|
|
r.y = 200
|
|
|
|
|
r.resize(300, 150)
|
|
|
|
|
return r.toJSON()
|
|
|
|
|
`)
|
|
|
|
|
expect(exitCode).toBe(0)
|
fix: validate parsed JSON at untrusted boundaries with Valibot (#855)
* fix: validate parsed JSON at untrusted boundaries with Valibot
Clipboard HTML, library revisions from shared storage, MCP and automation
WebSocket messages, the MCP discovery file, sidecar output and AI/MCP tool
arguments were JSON.parse'd and cast to their expected types, so a
malformed payload reached the document or crashed paste. They now go
through v.pipe(v.string(), v.parseJson(), Schema), which reports bad JSON
and a wrong shape as the same validation failure.
The path_set tool rejects an invalid VectorNetwork and shares its parser
with create_vector. The CLI library catalog validates its files and runs
revisions through the same size, identity and content-hash checks as the
app; reading image bytes as index-keyed records also stops them coming
back empty. Hand-rolled typeof readers for plugin data, document metadata,
caches and preferences become schemas with their behaviour preserved, and
readCacheJSON takes a schema for its payload.
open-pencil/no-unvalidated-json-parse rejects type assertions on
JSON.parse results other than `as unknown` in src and packages/*/src.
* refactor: validate parsed JSON in tests and tooling
Extend open-pencil/no-unvalidated-json-parse beyond source: tests, helpers and repo tooling now parse JSON through Valibot schemas instead of asserting a type. The shared fixture reader returns a validated object; its old array annotation never matched the fixtures.
* fix: validate clipboard geometry bytes, library images and model catalogs
Clipboard geometry blobs and library image bytes must be bytes at contiguous indexes, so out-of-range or gapped values are rejected instead of silently becoming different geometry or images; serialized library nodes must carry source metadata. The models.dev and OpenRouter responses are validated like their cached copies, and activate-tab rejects a CDP frame it cannot read instead of hanging.
* refactor: extend the JSON validation lint to .json() results
no-unvalidated-json-parse now also rejects type assertions on Response, Bun.file and shell .json() results, the same unchecked parse in another form. MCP server tests read /health through a validated readHealth helper and discovery files through parseDiscoveryInfo; the remaining tooling reads its JSON through schemas.
* test: validate the RPC request body in the CLI app export test
* test: validate CLI JSON output in the tool and app command tests
* test: compare the malformed models.dev fallback with the curated list
2026-10-04 17:01:50 +00:00
|
|
|
const result = parseRecord(stdout)
|
Unify tool definitions: define once, adapt for AI/CLI/MCP
Move tool logic to @open-pencil/core/tools/schema.ts as framework-agnostic
ToolDef objects. AI adapter generates valibot schemas + Vercel AI tool()
wrappers automatically.
26 tools (was 10): create_shape, render (JSX), set_fill, set_stroke,
set_effects, set_layout, set_constraints, update_node, delete, clone,
rename, reparent, group/ungroup, find_nodes, get_node, get_page_tree,
get_selection, select, list_pages, switch_page, list_variables,
list_collections, create_component, create_instance, eval.
src/ai/tools.ts: 269→28 lines (adapter only).
Tests for all 3 interfaces:
- 26 core tool tests (FigmaAPI directly)
- 11 AI adapter tests (valibot + Vercel AI SDK tool())
- 12 CLI integration tests (eval command on .fig fixture)
323 total, all passing.
2026-03-01 11:59:19 +00:00
|
|
|
expect(result.name).toBe('TestRect')
|
|
|
|
|
expect(result.x).toBe(100)
|
|
|
|
|
expect(result.y).toBe(200)
|
|
|
|
|
expect(result.width).toBe(300)
|
|
|
|
|
expect(result.height).toBe(150)
|
|
|
|
|
})
|
|
|
|
|
|
|
|
|
|
test('set fill on a node', async () => {
|
|
|
|
|
const { stdout, exitCode } = await evalCode(`
|
|
|
|
|
const r = figma.createRectangle()
|
|
|
|
|
r.resize(50, 50)
|
|
|
|
|
r.fills = [{ type: 'SOLID', color: { r: 1, g: 0, b: 0, a: 1 }, opacity: 1, visible: true }]
|
|
|
|
|
return { fills: r.fills }
|
|
|
|
|
`)
|
|
|
|
|
expect(exitCode).toBe(0)
|
fix: validate parsed JSON at untrusted boundaries with Valibot (#855)
* fix: validate parsed JSON at untrusted boundaries with Valibot
Clipboard HTML, library revisions from shared storage, MCP and automation
WebSocket messages, the MCP discovery file, sidecar output and AI/MCP tool
arguments were JSON.parse'd and cast to their expected types, so a
malformed payload reached the document or crashed paste. They now go
through v.pipe(v.string(), v.parseJson(), Schema), which reports bad JSON
and a wrong shape as the same validation failure.
The path_set tool rejects an invalid VectorNetwork and shares its parser
with create_vector. The CLI library catalog validates its files and runs
revisions through the same size, identity and content-hash checks as the
app; reading image bytes as index-keyed records also stops them coming
back empty. Hand-rolled typeof readers for plugin data, document metadata,
caches and preferences become schemas with their behaviour preserved, and
readCacheJSON takes a schema for its payload.
open-pencil/no-unvalidated-json-parse rejects type assertions on
JSON.parse results other than `as unknown` in src and packages/*/src.
* refactor: validate parsed JSON in tests and tooling
Extend open-pencil/no-unvalidated-json-parse beyond source: tests, helpers and repo tooling now parse JSON through Valibot schemas instead of asserting a type. The shared fixture reader returns a validated object; its old array annotation never matched the fixtures.
* fix: validate clipboard geometry bytes, library images and model catalogs
Clipboard geometry blobs and library image bytes must be bytes at contiguous indexes, so out-of-range or gapped values are rejected instead of silently becoming different geometry or images; serialized library nodes must carry source metadata. The models.dev and OpenRouter responses are validated like their cached copies, and activate-tab rejects a CDP frame it cannot read instead of hanging.
* refactor: extend the JSON validation lint to .json() results
no-unvalidated-json-parse now also rejects type assertions on Response, Bun.file and shell .json() results, the same unchecked parse in another form. MCP server tests read /health through a validated readHealth helper and discovery files through parseDiscoveryInfo; the remaining tooling reads its JSON through schemas.
* test: validate the RPC request body in the CLI app export test
* test: validate CLI JSON output in the tool and app command tests
* test: compare the malformed models.dev fallback with the curated list
2026-10-04 17:01:50 +00:00
|
|
|
const result = parseRecord(stdout)
|
test: typecheck the test suites and fix what that found (#896)
* build: typecheck the test suites
Tests were in no TypeScript program: no tsconfig included tests/** or
packages/*/tests/**, and bun strips types without checking them, so a
fixture could drop a required field and keep passing until something
read it.
@types/bun moves to the root because it was installed per package only,
and #cli-tests/* joins the paths the root config already carries.
* test: fix the type errors the test suites were hiding
Typechecking the tests turned up 1123 errors. Most were ordinary
strictness, but some were real: `NodeChange` bound to Figma's plugin
typings rather than the Kiwi codec in thirteen .fig tests,
materializeInstance was called with six arguments against five so the
blobs and source children were dropped, CanvasKit pixels were written
to a plain object that never reached WASM, and assertions were made
through accessors that do not exist, so they asserted nothing.
Fixtures that had quietly lost a required field now carry it, nullable
results are narrowed through the existing expectDefined helper rather
than assumed, and stand-ins for CanvasKit and the editor go through one
named helper instead of an unexplained cast at each site.
No test was deleted, skipped, or weakened, and no `any`, non-null
assertion, or ts-expect-error was introduced.
* docs: record what typechecking the tests established
Pins the app program's global types with an assertion rather than a
note, since an unpinned types list lets any root @types package decide
which platform src/** is judged against.
The two environment faults that look like code regressions — Vite's
dependency pre-bundle outliving a package rebuild, and heavy .fig
suites failing under load — go to the development docs, where an
explanation belongs.
* fix: align @types/bun and keep node types resolvable when extended
The root manifest declared a newer @types/bun than every package, which
check:monorepo rejects, and pinning the app program's types left them
unresolvable from a config that extends this one out of tree.
* fix: fail the test typecheck when the compiler itself fails
The gate matched diagnostics by substring, so a compiler or config
failure that named no test file printed a pass while having checked
nothing. Diagnostics are now split by whether they name a file: an
unscoped one is the run failing and stops the gate, a test file's is a
finding, and a source file's stays out by design.
Also drops the parameter planComponentConstruction never read, and
makes the inner-shadow verification script exit non-zero when it
renders no image instead of logging and succeeding.
* chore: merge master into tests-typecheck
2026-10-05 12:42:38 +00:00
|
|
|
const fills = v.parse(SolidFills, result.fills)
|
|
|
|
|
expect(fills.length).toBe(1)
|
|
|
|
|
expect(fills[0].color.r).toBe(1)
|
Unify tool definitions: define once, adapt for AI/CLI/MCP
Move tool logic to @open-pencil/core/tools/schema.ts as framework-agnostic
ToolDef objects. AI adapter generates valibot schemas + Vercel AI tool()
wrappers automatically.
26 tools (was 10): create_shape, render (JSX), set_fill, set_stroke,
set_effects, set_layout, set_constraints, update_node, delete, clone,
rename, reparent, group/ungroup, find_nodes, get_node, get_page_tree,
get_selection, select, list_pages, switch_page, list_variables,
list_collections, create_component, create_instance, eval.
src/ai/tools.ts: 269→28 lines (adapter only).
Tests for all 3 interfaces:
- 26 core tool tests (FigmaAPI directly)
- 11 AI adapter tests (valibot + Vercel AI SDK tool())
- 12 CLI integration tests (eval command on .fig fixture)
323 total, all passing.
2026-03-01 11:59:19 +00:00
|
|
|
})
|
|
|
|
|
|
|
|
|
|
test('set layout on a frame', async () => {
|
|
|
|
|
const { stdout, exitCode } = await evalCode(`
|
|
|
|
|
const f = figma.createFrame()
|
|
|
|
|
f.resize(300, 200)
|
|
|
|
|
f.layoutMode = 'VERTICAL'
|
|
|
|
|
f.itemSpacing = 16
|
|
|
|
|
f.paddingLeft = 20
|
|
|
|
|
f.paddingRight = 20
|
|
|
|
|
f.paddingTop = 20
|
|
|
|
|
f.paddingBottom = 20
|
|
|
|
|
return {
|
|
|
|
|
layoutMode: f.layoutMode,
|
|
|
|
|
itemSpacing: f.itemSpacing,
|
|
|
|
|
paddingLeft: f.paddingLeft
|
|
|
|
|
}
|
|
|
|
|
`)
|
|
|
|
|
expect(exitCode).toBe(0)
|
fix: validate parsed JSON at untrusted boundaries with Valibot (#855)
* fix: validate parsed JSON at untrusted boundaries with Valibot
Clipboard HTML, library revisions from shared storage, MCP and automation
WebSocket messages, the MCP discovery file, sidecar output and AI/MCP tool
arguments were JSON.parse'd and cast to their expected types, so a
malformed payload reached the document or crashed paste. They now go
through v.pipe(v.string(), v.parseJson(), Schema), which reports bad JSON
and a wrong shape as the same validation failure.
The path_set tool rejects an invalid VectorNetwork and shares its parser
with create_vector. The CLI library catalog validates its files and runs
revisions through the same size, identity and content-hash checks as the
app; reading image bytes as index-keyed records also stops them coming
back empty. Hand-rolled typeof readers for plugin data, document metadata,
caches and preferences become schemas with their behaviour preserved, and
readCacheJSON takes a schema for its payload.
open-pencil/no-unvalidated-json-parse rejects type assertions on
JSON.parse results other than `as unknown` in src and packages/*/src.
* refactor: validate parsed JSON in tests and tooling
Extend open-pencil/no-unvalidated-json-parse beyond source: tests, helpers and repo tooling now parse JSON through Valibot schemas instead of asserting a type. The shared fixture reader returns a validated object; its old array annotation never matched the fixtures.
* fix: validate clipboard geometry bytes, library images and model catalogs
Clipboard geometry blobs and library image bytes must be bytes at contiguous indexes, so out-of-range or gapped values are rejected instead of silently becoming different geometry or images; serialized library nodes must carry source metadata. The models.dev and OpenRouter responses are validated like their cached copies, and activate-tab rejects a CDP frame it cannot read instead of hanging.
* refactor: extend the JSON validation lint to .json() results
no-unvalidated-json-parse now also rejects type assertions on Response, Bun.file and shell .json() results, the same unchecked parse in another form. MCP server tests read /health through a validated readHealth helper and discovery files through parseDiscoveryInfo; the remaining tooling reads its JSON through schemas.
* test: validate the RPC request body in the CLI app export test
* test: validate CLI JSON output in the tool and app command tests
* test: compare the malformed models.dev fallback with the curated list
2026-10-04 17:01:50 +00:00
|
|
|
const result = parseRecord(stdout)
|
Unify tool definitions: define once, adapt for AI/CLI/MCP
Move tool logic to @open-pencil/core/tools/schema.ts as framework-agnostic
ToolDef objects. AI adapter generates valibot schemas + Vercel AI tool()
wrappers automatically.
26 tools (was 10): create_shape, render (JSX), set_fill, set_stroke,
set_effects, set_layout, set_constraints, update_node, delete, clone,
rename, reparent, group/ungroup, find_nodes, get_node, get_page_tree,
get_selection, select, list_pages, switch_page, list_variables,
list_collections, create_component, create_instance, eval.
src/ai/tools.ts: 269→28 lines (adapter only).
Tests for all 3 interfaces:
- 26 core tool tests (FigmaAPI directly)
- 11 AI adapter tests (valibot + Vercel AI SDK tool())
- 12 CLI integration tests (eval command on .fig fixture)
323 total, all passing.
2026-03-01 11:59:19 +00:00
|
|
|
expect(result.layoutMode).toBe('VERTICAL')
|
|
|
|
|
expect(result.itemSpacing).toBe(16)
|
|
|
|
|
expect(result.paddingLeft).toBe(20)
|
|
|
|
|
})
|
|
|
|
|
|
|
|
|
|
test('create component from node', async () => {
|
|
|
|
|
const { stdout, exitCode } = await evalCode(`
|
|
|
|
|
const f = figma.createFrame()
|
|
|
|
|
f.name = 'Button'
|
|
|
|
|
f.resize(200, 48)
|
|
|
|
|
const comp = figma.createComponentFromNode(f)
|
|
|
|
|
return { name: comp.name, type: comp.type }
|
|
|
|
|
`)
|
|
|
|
|
expect(exitCode).toBe(0)
|
fix: validate parsed JSON at untrusted boundaries with Valibot (#855)
* fix: validate parsed JSON at untrusted boundaries with Valibot
Clipboard HTML, library revisions from shared storage, MCP and automation
WebSocket messages, the MCP discovery file, sidecar output and AI/MCP tool
arguments were JSON.parse'd and cast to their expected types, so a
malformed payload reached the document or crashed paste. They now go
through v.pipe(v.string(), v.parseJson(), Schema), which reports bad JSON
and a wrong shape as the same validation failure.
The path_set tool rejects an invalid VectorNetwork and shares its parser
with create_vector. The CLI library catalog validates its files and runs
revisions through the same size, identity and content-hash checks as the
app; reading image bytes as index-keyed records also stops them coming
back empty. Hand-rolled typeof readers for plugin data, document metadata,
caches and preferences become schemas with their behaviour preserved, and
readCacheJSON takes a schema for its payload.
open-pencil/no-unvalidated-json-parse rejects type assertions on
JSON.parse results other than `as unknown` in src and packages/*/src.
* refactor: validate parsed JSON in tests and tooling
Extend open-pencil/no-unvalidated-json-parse beyond source: tests, helpers and repo tooling now parse JSON through Valibot schemas instead of asserting a type. The shared fixture reader returns a validated object; its old array annotation never matched the fixtures.
* fix: validate clipboard geometry bytes, library images and model catalogs
Clipboard geometry blobs and library image bytes must be bytes at contiguous indexes, so out-of-range or gapped values are rejected instead of silently becoming different geometry or images; serialized library nodes must carry source metadata. The models.dev and OpenRouter responses are validated like their cached copies, and activate-tab rejects a CDP frame it cannot read instead of hanging.
* refactor: extend the JSON validation lint to .json() results
no-unvalidated-json-parse now also rejects type assertions on Response, Bun.file and shell .json() results, the same unchecked parse in another form. MCP server tests read /health through a validated readHealth helper and discovery files through parseDiscoveryInfo; the remaining tooling reads its JSON through schemas.
* test: validate the RPC request body in the CLI app export test
* test: validate CLI JSON output in the tool and app command tests
* test: compare the malformed models.dev fallback with the curated list
2026-10-04 17:01:50 +00:00
|
|
|
const result = parseRecord(stdout)
|
Unify tool definitions: define once, adapt for AI/CLI/MCP
Move tool logic to @open-pencil/core/tools/schema.ts as framework-agnostic
ToolDef objects. AI adapter generates valibot schemas + Vercel AI tool()
wrappers automatically.
26 tools (was 10): create_shape, render (JSX), set_fill, set_stroke,
set_effects, set_layout, set_constraints, update_node, delete, clone,
rename, reparent, group/ungroup, find_nodes, get_node, get_page_tree,
get_selection, select, list_pages, switch_page, list_variables,
list_collections, create_component, create_instance, eval.
src/ai/tools.ts: 269→28 lines (adapter only).
Tests for all 3 interfaces:
- 26 core tool tests (FigmaAPI directly)
- 11 AI adapter tests (valibot + Vercel AI SDK tool())
- 12 CLI integration tests (eval command on .fig fixture)
323 total, all passing.
2026-03-01 11:59:19 +00:00
|
|
|
expect(result.name).toBe('Button')
|
|
|
|
|
expect(result.type).toBe('COMPONENT')
|
|
|
|
|
})
|
|
|
|
|
|
|
|
|
|
test('group and ungroup nodes', async () => {
|
|
|
|
|
const { stdout, exitCode } = await evalCode(`
|
|
|
|
|
const r1 = figma.createRectangle()
|
|
|
|
|
r1.resize(50, 50)
|
|
|
|
|
const r2 = figma.createRectangle()
|
|
|
|
|
r2.resize(50, 50)
|
|
|
|
|
const group = figma.group([r1, r2], figma.currentPage)
|
|
|
|
|
const groupType = group.type
|
|
|
|
|
const childCount = group.children.length
|
|
|
|
|
figma.ungroup(group)
|
|
|
|
|
const ungrouped = figma.getNodeById(group.id)
|
|
|
|
|
return { groupType, childCount, ungroupedExists: ungrouped !== null }
|
|
|
|
|
`)
|
|
|
|
|
expect(exitCode).toBe(0)
|
fix: validate parsed JSON at untrusted boundaries with Valibot (#855)
* fix: validate parsed JSON at untrusted boundaries with Valibot
Clipboard HTML, library revisions from shared storage, MCP and automation
WebSocket messages, the MCP discovery file, sidecar output and AI/MCP tool
arguments were JSON.parse'd and cast to their expected types, so a
malformed payload reached the document or crashed paste. They now go
through v.pipe(v.string(), v.parseJson(), Schema), which reports bad JSON
and a wrong shape as the same validation failure.
The path_set tool rejects an invalid VectorNetwork and shares its parser
with create_vector. The CLI library catalog validates its files and runs
revisions through the same size, identity and content-hash checks as the
app; reading image bytes as index-keyed records also stops them coming
back empty. Hand-rolled typeof readers for plugin data, document metadata,
caches and preferences become schemas with their behaviour preserved, and
readCacheJSON takes a schema for its payload.
open-pencil/no-unvalidated-json-parse rejects type assertions on
JSON.parse results other than `as unknown` in src and packages/*/src.
* refactor: validate parsed JSON in tests and tooling
Extend open-pencil/no-unvalidated-json-parse beyond source: tests, helpers and repo tooling now parse JSON through Valibot schemas instead of asserting a type. The shared fixture reader returns a validated object; its old array annotation never matched the fixtures.
* fix: validate clipboard geometry bytes, library images and model catalogs
Clipboard geometry blobs and library image bytes must be bytes at contiguous indexes, so out-of-range or gapped values are rejected instead of silently becoming different geometry or images; serialized library nodes must carry source metadata. The models.dev and OpenRouter responses are validated like their cached copies, and activate-tab rejects a CDP frame it cannot read instead of hanging.
* refactor: extend the JSON validation lint to .json() results
no-unvalidated-json-parse now also rejects type assertions on Response, Bun.file and shell .json() results, the same unchecked parse in another form. MCP server tests read /health through a validated readHealth helper and discovery files through parseDiscoveryInfo; the remaining tooling reads its JSON through schemas.
* test: validate the RPC request body in the CLI app export test
* test: validate CLI JSON output in the tool and app command tests
* test: compare the malformed models.dev fallback with the curated list
2026-10-04 17:01:50 +00:00
|
|
|
const result = parseRecord(stdout)
|
Unify tool definitions: define once, adapt for AI/CLI/MCP
Move tool logic to @open-pencil/core/tools/schema.ts as framework-agnostic
ToolDef objects. AI adapter generates valibot schemas + Vercel AI tool()
wrappers automatically.
26 tools (was 10): create_shape, render (JSX), set_fill, set_stroke,
set_effects, set_layout, set_constraints, update_node, delete, clone,
rename, reparent, group/ungroup, find_nodes, get_node, get_page_tree,
get_selection, select, list_pages, switch_page, list_variables,
list_collections, create_component, create_instance, eval.
src/ai/tools.ts: 269→28 lines (adapter only).
Tests for all 3 interfaces:
- 26 core tool tests (FigmaAPI directly)
- 11 AI adapter tests (valibot + Vercel AI SDK tool())
- 12 CLI integration tests (eval command on .fig fixture)
323 total, all passing.
2026-03-01 11:59:19 +00:00
|
|
|
expect(result.groupType).toBe('GROUP')
|
|
|
|
|
expect(result.childCount).toBe(2)
|
|
|
|
|
expect(result.ungroupedExists).toBe(false)
|
|
|
|
|
})
|
|
|
|
|
|
|
|
|
|
test('find nodes by type on fixture', async () => {
|
|
|
|
|
const { stdout, exitCode } = await evalCode(`
|
|
|
|
|
const texts = figma.currentPage.findAllWithCriteria({ types: ['TEXT'] })
|
|
|
|
|
return { count: texts.length, hasTexts: texts.length > 0 }
|
|
|
|
|
`)
|
|
|
|
|
expect(exitCode).toBe(0)
|
fix: validate parsed JSON at untrusted boundaries with Valibot (#855)
* fix: validate parsed JSON at untrusted boundaries with Valibot
Clipboard HTML, library revisions from shared storage, MCP and automation
WebSocket messages, the MCP discovery file, sidecar output and AI/MCP tool
arguments were JSON.parse'd and cast to their expected types, so a
malformed payload reached the document or crashed paste. They now go
through v.pipe(v.string(), v.parseJson(), Schema), which reports bad JSON
and a wrong shape as the same validation failure.
The path_set tool rejects an invalid VectorNetwork and shares its parser
with create_vector. The CLI library catalog validates its files and runs
revisions through the same size, identity and content-hash checks as the
app; reading image bytes as index-keyed records also stops them coming
back empty. Hand-rolled typeof readers for plugin data, document metadata,
caches and preferences become schemas with their behaviour preserved, and
readCacheJSON takes a schema for its payload.
open-pencil/no-unvalidated-json-parse rejects type assertions on
JSON.parse results other than `as unknown` in src and packages/*/src.
* refactor: validate parsed JSON in tests and tooling
Extend open-pencil/no-unvalidated-json-parse beyond source: tests, helpers and repo tooling now parse JSON through Valibot schemas instead of asserting a type. The shared fixture reader returns a validated object; its old array annotation never matched the fixtures.
* fix: validate clipboard geometry bytes, library images and model catalogs
Clipboard geometry blobs and library image bytes must be bytes at contiguous indexes, so out-of-range or gapped values are rejected instead of silently becoming different geometry or images; serialized library nodes must carry source metadata. The models.dev and OpenRouter responses are validated like their cached copies, and activate-tab rejects a CDP frame it cannot read instead of hanging.
* refactor: extend the JSON validation lint to .json() results
no-unvalidated-json-parse now also rejects type assertions on Response, Bun.file and shell .json() results, the same unchecked parse in another form. MCP server tests read /health through a validated readHealth helper and discovery files through parseDiscoveryInfo; the remaining tooling reads its JSON through schemas.
* test: validate the RPC request body in the CLI app export test
* test: validate CLI JSON output in the tool and app command tests
* test: compare the malformed models.dev fallback with the curated list
2026-10-04 17:01:50 +00:00
|
|
|
const result = parseRecord(stdout)
|
Unify tool definitions: define once, adapt for AI/CLI/MCP
Move tool logic to @open-pencil/core/tools/schema.ts as framework-agnostic
ToolDef objects. AI adapter generates valibot schemas + Vercel AI tool()
wrappers automatically.
26 tools (was 10): create_shape, render (JSX), set_fill, set_stroke,
set_effects, set_layout, set_constraints, update_node, delete, clone,
rename, reparent, group/ungroup, find_nodes, get_node, get_page_tree,
get_selection, select, list_pages, switch_page, list_variables,
list_collections, create_component, create_instance, eval.
src/ai/tools.ts: 269→28 lines (adapter only).
Tests for all 3 interfaces:
- 26 core tool tests (FigmaAPI directly)
- 11 AI adapter tests (valibot + Vercel AI SDK tool())
- 12 CLI integration tests (eval command on .fig fixture)
323 total, all passing.
2026-03-01 11:59:19 +00:00
|
|
|
expect(result.hasTexts).toBe(true)
|
|
|
|
|
expect(result.count).toBeGreaterThan(0)
|
|
|
|
|
})
|
|
|
|
|
|
|
|
|
|
test('clone a node', async () => {
|
|
|
|
|
const { stdout, exitCode } = await evalCode(`
|
|
|
|
|
const r = figma.createRectangle()
|
|
|
|
|
r.name = 'Original'
|
|
|
|
|
r.resize(100, 100)
|
|
|
|
|
const clone = r.clone()
|
|
|
|
|
return {
|
|
|
|
|
same: r.id === clone.id,
|
|
|
|
|
cloneName: clone.name,
|
|
|
|
|
cloneWidth: clone.width
|
|
|
|
|
}
|
|
|
|
|
`)
|
|
|
|
|
expect(exitCode).toBe(0)
|
fix: validate parsed JSON at untrusted boundaries with Valibot (#855)
* fix: validate parsed JSON at untrusted boundaries with Valibot
Clipboard HTML, library revisions from shared storage, MCP and automation
WebSocket messages, the MCP discovery file, sidecar output and AI/MCP tool
arguments were JSON.parse'd and cast to their expected types, so a
malformed payload reached the document or crashed paste. They now go
through v.pipe(v.string(), v.parseJson(), Schema), which reports bad JSON
and a wrong shape as the same validation failure.
The path_set tool rejects an invalid VectorNetwork and shares its parser
with create_vector. The CLI library catalog validates its files and runs
revisions through the same size, identity and content-hash checks as the
app; reading image bytes as index-keyed records also stops them coming
back empty. Hand-rolled typeof readers for plugin data, document metadata,
caches and preferences become schemas with their behaviour preserved, and
readCacheJSON takes a schema for its payload.
open-pencil/no-unvalidated-json-parse rejects type assertions on
JSON.parse results other than `as unknown` in src and packages/*/src.
* refactor: validate parsed JSON in tests and tooling
Extend open-pencil/no-unvalidated-json-parse beyond source: tests, helpers and repo tooling now parse JSON through Valibot schemas instead of asserting a type. The shared fixture reader returns a validated object; its old array annotation never matched the fixtures.
* fix: validate clipboard geometry bytes, library images and model catalogs
Clipboard geometry blobs and library image bytes must be bytes at contiguous indexes, so out-of-range or gapped values are rejected instead of silently becoming different geometry or images; serialized library nodes must carry source metadata. The models.dev and OpenRouter responses are validated like their cached copies, and activate-tab rejects a CDP frame it cannot read instead of hanging.
* refactor: extend the JSON validation lint to .json() results
no-unvalidated-json-parse now also rejects type assertions on Response, Bun.file and shell .json() results, the same unchecked parse in another form. MCP server tests read /health through a validated readHealth helper and discovery files through parseDiscoveryInfo; the remaining tooling reads its JSON through schemas.
* test: validate the RPC request body in the CLI app export test
* test: validate CLI JSON output in the tool and app command tests
* test: compare the malformed models.dev fallback with the curated list
2026-10-04 17:01:50 +00:00
|
|
|
const result = parseRecord(stdout)
|
Unify tool definitions: define once, adapt for AI/CLI/MCP
Move tool logic to @open-pencil/core/tools/schema.ts as framework-agnostic
ToolDef objects. AI adapter generates valibot schemas + Vercel AI tool()
wrappers automatically.
26 tools (was 10): create_shape, render (JSX), set_fill, set_stroke,
set_effects, set_layout, set_constraints, update_node, delete, clone,
rename, reparent, group/ungroup, find_nodes, get_node, get_page_tree,
get_selection, select, list_pages, switch_page, list_variables,
list_collections, create_component, create_instance, eval.
src/ai/tools.ts: 269→28 lines (adapter only).
Tests for all 3 interfaces:
- 26 core tool tests (FigmaAPI directly)
- 11 AI adapter tests (valibot + Vercel AI SDK tool())
- 12 CLI integration tests (eval command on .fig fixture)
323 total, all passing.
2026-03-01 11:59:19 +00:00
|
|
|
expect(result.same).toBe(false)
|
|
|
|
|
expect(result.cloneName).toBe('Original')
|
|
|
|
|
expect(result.cloneWidth).toBe(100)
|
|
|
|
|
})
|
|
|
|
|
|
|
|
|
|
test('reparent node into frame', async () => {
|
|
|
|
|
const { stdout, exitCode } = await evalCode(`
|
|
|
|
|
const frame = figma.createFrame()
|
|
|
|
|
frame.resize(300, 300)
|
|
|
|
|
const rect = figma.createRectangle()
|
|
|
|
|
rect.resize(50, 50)
|
|
|
|
|
frame.appendChild(rect)
|
|
|
|
|
return {
|
|
|
|
|
parentId: rect.parent?.id,
|
|
|
|
|
isChild: frame.children.some(c => c.id === rect.id)
|
|
|
|
|
}
|
|
|
|
|
`)
|
|
|
|
|
expect(exitCode).toBe(0)
|
fix: validate parsed JSON at untrusted boundaries with Valibot (#855)
* fix: validate parsed JSON at untrusted boundaries with Valibot
Clipboard HTML, library revisions from shared storage, MCP and automation
WebSocket messages, the MCP discovery file, sidecar output and AI/MCP tool
arguments were JSON.parse'd and cast to their expected types, so a
malformed payload reached the document or crashed paste. They now go
through v.pipe(v.string(), v.parseJson(), Schema), which reports bad JSON
and a wrong shape as the same validation failure.
The path_set tool rejects an invalid VectorNetwork and shares its parser
with create_vector. The CLI library catalog validates its files and runs
revisions through the same size, identity and content-hash checks as the
app; reading image bytes as index-keyed records also stops them coming
back empty. Hand-rolled typeof readers for plugin data, document metadata,
caches and preferences become schemas with their behaviour preserved, and
readCacheJSON takes a schema for its payload.
open-pencil/no-unvalidated-json-parse rejects type assertions on
JSON.parse results other than `as unknown` in src and packages/*/src.
* refactor: validate parsed JSON in tests and tooling
Extend open-pencil/no-unvalidated-json-parse beyond source: tests, helpers and repo tooling now parse JSON through Valibot schemas instead of asserting a type. The shared fixture reader returns a validated object; its old array annotation never matched the fixtures.
* fix: validate clipboard geometry bytes, library images and model catalogs
Clipboard geometry blobs and library image bytes must be bytes at contiguous indexes, so out-of-range or gapped values are rejected instead of silently becoming different geometry or images; serialized library nodes must carry source metadata. The models.dev and OpenRouter responses are validated like their cached copies, and activate-tab rejects a CDP frame it cannot read instead of hanging.
* refactor: extend the JSON validation lint to .json() results
no-unvalidated-json-parse now also rejects type assertions on Response, Bun.file and shell .json() results, the same unchecked parse in another form. MCP server tests read /health through a validated readHealth helper and discovery files through parseDiscoveryInfo; the remaining tooling reads its JSON through schemas.
* test: validate the RPC request body in the CLI app export test
* test: validate CLI JSON output in the tool and app command tests
* test: compare the malformed models.dev fallback with the curated list
2026-10-04 17:01:50 +00:00
|
|
|
const result = parseRecord(stdout)
|
Unify tool definitions: define once, adapt for AI/CLI/MCP
Move tool logic to @open-pencil/core/tools/schema.ts as framework-agnostic
ToolDef objects. AI adapter generates valibot schemas + Vercel AI tool()
wrappers automatically.
26 tools (was 10): create_shape, render (JSX), set_fill, set_stroke,
set_effects, set_layout, set_constraints, update_node, delete, clone,
rename, reparent, group/ungroup, find_nodes, get_node, get_page_tree,
get_selection, select, list_pages, switch_page, list_variables,
list_collections, create_component, create_instance, eval.
src/ai/tools.ts: 269→28 lines (adapter only).
Tests for all 3 interfaces:
- 26 core tool tests (FigmaAPI directly)
- 11 AI adapter tests (valibot + Vercel AI SDK tool())
- 12 CLI integration tests (eval command on .fig fixture)
323 total, all passing.
2026-03-01 11:59:19 +00:00
|
|
|
expect(result.isChild).toBe(true)
|
|
|
|
|
})
|
|
|
|
|
|
|
|
|
|
test('set constraints', async () => {
|
|
|
|
|
const { stdout, exitCode } = await evalCode(`
|
|
|
|
|
const r = figma.createRectangle()
|
|
|
|
|
r.resize(100, 100)
|
|
|
|
|
r.constraints = { horizontal: 'CENTER', vertical: 'STRETCH' }
|
|
|
|
|
return r.constraints
|
|
|
|
|
`)
|
|
|
|
|
expect(exitCode).toBe(0)
|
fix: validate parsed JSON at untrusted boundaries with Valibot (#855)
* fix: validate parsed JSON at untrusted boundaries with Valibot
Clipboard HTML, library revisions from shared storage, MCP and automation
WebSocket messages, the MCP discovery file, sidecar output and AI/MCP tool
arguments were JSON.parse'd and cast to their expected types, so a
malformed payload reached the document or crashed paste. They now go
through v.pipe(v.string(), v.parseJson(), Schema), which reports bad JSON
and a wrong shape as the same validation failure.
The path_set tool rejects an invalid VectorNetwork and shares its parser
with create_vector. The CLI library catalog validates its files and runs
revisions through the same size, identity and content-hash checks as the
app; reading image bytes as index-keyed records also stops them coming
back empty. Hand-rolled typeof readers for plugin data, document metadata,
caches and preferences become schemas with their behaviour preserved, and
readCacheJSON takes a schema for its payload.
open-pencil/no-unvalidated-json-parse rejects type assertions on
JSON.parse results other than `as unknown` in src and packages/*/src.
* refactor: validate parsed JSON in tests and tooling
Extend open-pencil/no-unvalidated-json-parse beyond source: tests, helpers and repo tooling now parse JSON through Valibot schemas instead of asserting a type. The shared fixture reader returns a validated object; its old array annotation never matched the fixtures.
* fix: validate clipboard geometry bytes, library images and model catalogs
Clipboard geometry blobs and library image bytes must be bytes at contiguous indexes, so out-of-range or gapped values are rejected instead of silently becoming different geometry or images; serialized library nodes must carry source metadata. The models.dev and OpenRouter responses are validated like their cached copies, and activate-tab rejects a CDP frame it cannot read instead of hanging.
* refactor: extend the JSON validation lint to .json() results
no-unvalidated-json-parse now also rejects type assertions on Response, Bun.file and shell .json() results, the same unchecked parse in another form. MCP server tests read /health through a validated readHealth helper and discovery files through parseDiscoveryInfo; the remaining tooling reads its JSON through schemas.
* test: validate the RPC request body in the CLI app export test
* test: validate CLI JSON output in the tool and app command tests
* test: compare the malformed models.dev fallback with the curated list
2026-10-04 17:01:50 +00:00
|
|
|
const result = parseRecord(stdout)
|
Unify tool definitions: define once, adapt for AI/CLI/MCP
Move tool logic to @open-pencil/core/tools/schema.ts as framework-agnostic
ToolDef objects. AI adapter generates valibot schemas + Vercel AI tool()
wrappers automatically.
26 tools (was 10): create_shape, render (JSX), set_fill, set_stroke,
set_effects, set_layout, set_constraints, update_node, delete, clone,
rename, reparent, group/ungroup, find_nodes, get_node, get_page_tree,
get_selection, select, list_pages, switch_page, list_variables,
list_collections, create_component, create_instance, eval.
src/ai/tools.ts: 269→28 lines (adapter only).
Tests for all 3 interfaces:
- 26 core tool tests (FigmaAPI directly)
- 11 AI adapter tests (valibot + Vercel AI SDK tool())
- 12 CLI integration tests (eval command on .fig fixture)
323 total, all passing.
2026-03-01 11:59:19 +00:00
|
|
|
expect(result.horizontal).toBe('CENTER')
|
|
|
|
|
expect(result.vertical).toBe('STRETCH')
|
|
|
|
|
})
|
|
|
|
|
|
|
|
|
|
test('set effects', async () => {
|
|
|
|
|
const { stdout, exitCode } = await evalCode(`
|
|
|
|
|
const f = figma.createFrame()
|
|
|
|
|
f.resize(100, 100)
|
|
|
|
|
f.effects = [{
|
|
|
|
|
type: 'DROP_SHADOW',
|
|
|
|
|
color: { r: 0, g: 0, b: 0, a: 0.25 },
|
|
|
|
|
offset: { x: 0, y: 4 },
|
|
|
|
|
radius: 8,
|
|
|
|
|
spread: 0,
|
fix(figma-api): validate effects like Figma (#794)
* build(core): import Markdown with unplugin-raw
Core inlined ?raw imports with a hand-written Rolldown plugin that also turned plain .md imports into strings, which nothing in Core uses. unplugin-raw already does this for the Vue SDK and design-jsx; use it here too and drop the unused *.md module declaration.
* refactor(pen): use the scene-graph color parser
pen/src/color.ts duplicated parseColor from @open-pencil/scene-graph/color line for line. Import it instead, which also drops pen's direct culori dependency.
* fix(figma-api): validate effects like Figma
The effects setter stored whatever a script passed, so scripts that Figma rejects ran here and malformed effects reached rendering and .fig export. Validate against Figma's effect shapes with Valibot, recorded from live Figma: strict objects, required shadow fields, radius >= 0, RGBA channels within 0..1, and no shadow fields on blurs. The getter now returns Figma's shape so node.effects = node.effects keeps working.
Closes #786
* fix(figma-api): reject infinite numbers in effects
Figma rejects Infinity in every effect number ("Number must be finite"), but v.number() accepts it, so infinite radii, offsets, and spreads reached the scene graph.
* fix(figma-api): store PASS_THROUGH shadow blend as NORMAL
PASS_THROUGH is a layer blend mode. Live Figma accepts it on drop and inner shadows but reads NORMAL back, so do the same instead of storing it.
2026-09-30 17:18:20 +00:00
|
|
|
visible: true,
|
|
|
|
|
blendMode: 'NORMAL'
|
Unify tool definitions: define once, adapt for AI/CLI/MCP
Move tool logic to @open-pencil/core/tools/schema.ts as framework-agnostic
ToolDef objects. AI adapter generates valibot schemas + Vercel AI tool()
wrappers automatically.
26 tools (was 10): create_shape, render (JSX), set_fill, set_stroke,
set_effects, set_layout, set_constraints, update_node, delete, clone,
rename, reparent, group/ungroup, find_nodes, get_node, get_page_tree,
get_selection, select, list_pages, switch_page, list_variables,
list_collections, create_component, create_instance, eval.
src/ai/tools.ts: 269→28 lines (adapter only).
Tests for all 3 interfaces:
- 26 core tool tests (FigmaAPI directly)
- 11 AI adapter tests (valibot + Vercel AI SDK tool())
- 12 CLI integration tests (eval command on .fig fixture)
323 total, all passing.
2026-03-01 11:59:19 +00:00
|
|
|
}]
|
|
|
|
|
return { count: f.effects.length, type: f.effects[0].type }
|
|
|
|
|
`)
|
|
|
|
|
expect(exitCode).toBe(0)
|
fix: validate parsed JSON at untrusted boundaries with Valibot (#855)
* fix: validate parsed JSON at untrusted boundaries with Valibot
Clipboard HTML, library revisions from shared storage, MCP and automation
WebSocket messages, the MCP discovery file, sidecar output and AI/MCP tool
arguments were JSON.parse'd and cast to their expected types, so a
malformed payload reached the document or crashed paste. They now go
through v.pipe(v.string(), v.parseJson(), Schema), which reports bad JSON
and a wrong shape as the same validation failure.
The path_set tool rejects an invalid VectorNetwork and shares its parser
with create_vector. The CLI library catalog validates its files and runs
revisions through the same size, identity and content-hash checks as the
app; reading image bytes as index-keyed records also stops them coming
back empty. Hand-rolled typeof readers for plugin data, document metadata,
caches and preferences become schemas with their behaviour preserved, and
readCacheJSON takes a schema for its payload.
open-pencil/no-unvalidated-json-parse rejects type assertions on
JSON.parse results other than `as unknown` in src and packages/*/src.
* refactor: validate parsed JSON in tests and tooling
Extend open-pencil/no-unvalidated-json-parse beyond source: tests, helpers and repo tooling now parse JSON through Valibot schemas instead of asserting a type. The shared fixture reader returns a validated object; its old array annotation never matched the fixtures.
* fix: validate clipboard geometry bytes, library images and model catalogs
Clipboard geometry blobs and library image bytes must be bytes at contiguous indexes, so out-of-range or gapped values are rejected instead of silently becoming different geometry or images; serialized library nodes must carry source metadata. The models.dev and OpenRouter responses are validated like their cached copies, and activate-tab rejects a CDP frame it cannot read instead of hanging.
* refactor: extend the JSON validation lint to .json() results
no-unvalidated-json-parse now also rejects type assertions on Response, Bun.file and shell .json() results, the same unchecked parse in another form. MCP server tests read /health through a validated readHealth helper and discovery files through parseDiscoveryInfo; the remaining tooling reads its JSON through schemas.
* test: validate the RPC request body in the CLI app export test
* test: validate CLI JSON output in the tool and app command tests
* test: compare the malformed models.dev fallback with the curated list
2026-10-04 17:01:50 +00:00
|
|
|
const result = parseRecord(stdout)
|
Unify tool definitions: define once, adapt for AI/CLI/MCP
Move tool logic to @open-pencil/core/tools/schema.ts as framework-agnostic
ToolDef objects. AI adapter generates valibot schemas + Vercel AI tool()
wrappers automatically.
26 tools (was 10): create_shape, render (JSX), set_fill, set_stroke,
set_effects, set_layout, set_constraints, update_node, delete, clone,
rename, reparent, group/ungroup, find_nodes, get_node, get_page_tree,
get_selection, select, list_pages, switch_page, list_variables,
list_collections, create_component, create_instance, eval.
src/ai/tools.ts: 269→28 lines (adapter only).
Tests for all 3 interfaces:
- 26 core tool tests (FigmaAPI directly)
- 11 AI adapter tests (valibot + Vercel AI SDK tool())
- 12 CLI integration tests (eval command on .fig fixture)
323 total, all passing.
2026-03-01 11:59:19 +00:00
|
|
|
expect(result.count).toBe(1)
|
|
|
|
|
expect(result.type).toBe('DROP_SHADOW')
|
|
|
|
|
})
|
|
|
|
|
|
|
|
|
|
test('list variables from fixture', async () => {
|
|
|
|
|
const { stdout, exitCode } = await evalCode(`
|
|
|
|
|
const vars = figma.getLocalVariables()
|
|
|
|
|
const cols = figma.getLocalVariableCollections()
|
|
|
|
|
return { variables: vars.length, collections: cols.length }
|
|
|
|
|
`)
|
|
|
|
|
expect(exitCode).toBe(0)
|
fix: validate parsed JSON at untrusted boundaries with Valibot (#855)
* fix: validate parsed JSON at untrusted boundaries with Valibot
Clipboard HTML, library revisions from shared storage, MCP and automation
WebSocket messages, the MCP discovery file, sidecar output and AI/MCP tool
arguments were JSON.parse'd and cast to their expected types, so a
malformed payload reached the document or crashed paste. They now go
through v.pipe(v.string(), v.parseJson(), Schema), which reports bad JSON
and a wrong shape as the same validation failure.
The path_set tool rejects an invalid VectorNetwork and shares its parser
with create_vector. The CLI library catalog validates its files and runs
revisions through the same size, identity and content-hash checks as the
app; reading image bytes as index-keyed records also stops them coming
back empty. Hand-rolled typeof readers for plugin data, document metadata,
caches and preferences become schemas with their behaviour preserved, and
readCacheJSON takes a schema for its payload.
open-pencil/no-unvalidated-json-parse rejects type assertions on
JSON.parse results other than `as unknown` in src and packages/*/src.
* refactor: validate parsed JSON in tests and tooling
Extend open-pencil/no-unvalidated-json-parse beyond source: tests, helpers and repo tooling now parse JSON through Valibot schemas instead of asserting a type. The shared fixture reader returns a validated object; its old array annotation never matched the fixtures.
* fix: validate clipboard geometry bytes, library images and model catalogs
Clipboard geometry blobs and library image bytes must be bytes at contiguous indexes, so out-of-range or gapped values are rejected instead of silently becoming different geometry or images; serialized library nodes must carry source metadata. The models.dev and OpenRouter responses are validated like their cached copies, and activate-tab rejects a CDP frame it cannot read instead of hanging.
* refactor: extend the JSON validation lint to .json() results
no-unvalidated-json-parse now also rejects type assertions on Response, Bun.file and shell .json() results, the same unchecked parse in another form. MCP server tests read /health through a validated readHealth helper and discovery files through parseDiscoveryInfo; the remaining tooling reads its JSON through schemas.
* test: validate the RPC request body in the CLI app export test
* test: validate CLI JSON output in the tool and app command tests
* test: compare the malformed models.dev fallback with the curated list
2026-10-04 17:01:50 +00:00
|
|
|
const result = parseRecord(stdout)
|
Unify tool definitions: define once, adapt for AI/CLI/MCP
Move tool logic to @open-pencil/core/tools/schema.ts as framework-agnostic
ToolDef objects. AI adapter generates valibot schemas + Vercel AI tool()
wrappers automatically.
26 tools (was 10): create_shape, render (JSX), set_fill, set_stroke,
set_effects, set_layout, set_constraints, update_node, delete, clone,
rename, reparent, group/ungroup, find_nodes, get_node, get_page_tree,
get_selection, select, list_pages, switch_page, list_variables,
list_collections, create_component, create_instance, eval.
src/ai/tools.ts: 269→28 lines (adapter only).
Tests for all 3 interfaces:
- 26 core tool tests (FigmaAPI directly)
- 11 AI adapter tests (valibot + Vercel AI SDK tool())
- 12 CLI integration tests (eval command on .fig fixture)
323 total, all passing.
2026-03-01 11:59:19 +00:00
|
|
|
expect(typeof result.variables).toBe('number')
|
|
|
|
|
expect(typeof result.collections).toBe('number')
|
|
|
|
|
})
|
|
|
|
|
|
|
|
|
|
test('switch page', async () => {
|
|
|
|
|
const { stdout, exitCode } = await evalCode(`
|
|
|
|
|
const pages = figma.root.children
|
|
|
|
|
const first = pages[0]
|
|
|
|
|
figma.currentPage = first
|
|
|
|
|
return { page: figma.currentPage.name, pageCount: pages.length }
|
|
|
|
|
`)
|
|
|
|
|
expect(exitCode).toBe(0)
|
fix: validate parsed JSON at untrusted boundaries with Valibot (#855)
* fix: validate parsed JSON at untrusted boundaries with Valibot
Clipboard HTML, library revisions from shared storage, MCP and automation
WebSocket messages, the MCP discovery file, sidecar output and AI/MCP tool
arguments were JSON.parse'd and cast to their expected types, so a
malformed payload reached the document or crashed paste. They now go
through v.pipe(v.string(), v.parseJson(), Schema), which reports bad JSON
and a wrong shape as the same validation failure.
The path_set tool rejects an invalid VectorNetwork and shares its parser
with create_vector. The CLI library catalog validates its files and runs
revisions through the same size, identity and content-hash checks as the
app; reading image bytes as index-keyed records also stops them coming
back empty. Hand-rolled typeof readers for plugin data, document metadata,
caches and preferences become schemas with their behaviour preserved, and
readCacheJSON takes a schema for its payload.
open-pencil/no-unvalidated-json-parse rejects type assertions on
JSON.parse results other than `as unknown` in src and packages/*/src.
* refactor: validate parsed JSON in tests and tooling
Extend open-pencil/no-unvalidated-json-parse beyond source: tests, helpers and repo tooling now parse JSON through Valibot schemas instead of asserting a type. The shared fixture reader returns a validated object; its old array annotation never matched the fixtures.
* fix: validate clipboard geometry bytes, library images and model catalogs
Clipboard geometry blobs and library image bytes must be bytes at contiguous indexes, so out-of-range or gapped values are rejected instead of silently becoming different geometry or images; serialized library nodes must carry source metadata. The models.dev and OpenRouter responses are validated like their cached copies, and activate-tab rejects a CDP frame it cannot read instead of hanging.
* refactor: extend the JSON validation lint to .json() results
no-unvalidated-json-parse now also rejects type assertions on Response, Bun.file and shell .json() results, the same unchecked parse in another form. MCP server tests read /health through a validated readHealth helper and discovery files through parseDiscoveryInfo; the remaining tooling reads its JSON through schemas.
* test: validate the RPC request body in the CLI app export test
* test: validate CLI JSON output in the tool and app command tests
* test: compare the malformed models.dev fallback with the curated list
2026-10-04 17:01:50 +00:00
|
|
|
const result = parseRecord(stdout)
|
Unify tool definitions: define once, adapt for AI/CLI/MCP
Move tool logic to @open-pencil/core/tools/schema.ts as framework-agnostic
ToolDef objects. AI adapter generates valibot schemas + Vercel AI tool()
wrappers automatically.
26 tools (was 10): create_shape, render (JSX), set_fill, set_stroke,
set_effects, set_layout, set_constraints, update_node, delete, clone,
rename, reparent, group/ungroup, find_nodes, get_node, get_page_tree,
get_selection, select, list_pages, switch_page, list_variables,
list_collections, create_component, create_instance, eval.
src/ai/tools.ts: 269→28 lines (adapter only).
Tests for all 3 interfaces:
- 26 core tool tests (FigmaAPI directly)
- 11 AI adapter tests (valibot + Vercel AI SDK tool())
- 12 CLI integration tests (eval command on .fig fixture)
323 total, all passing.
2026-03-01 11:59:19 +00:00
|
|
|
expect(result.page).toBeTruthy()
|
|
|
|
|
expect(result.pageCount).toBeGreaterThanOrEqual(1)
|
|
|
|
|
})
|
|
|
|
|
})
|