Run `corepack use pnpm@next-12` (resolved to 12.3.4) on every directory
with a package.json: the repo root, the 11 module workspaces, and all
submodules. Every packageManager field now carries the same pinned
version and hash; the root and backend move off 11.20.0.
Fix the composable-test-suite workspace config (esbuild allowBuilds
placeholder left by pnpm 12) so its install passes, and add the missing
packageManager fields to frontend/packages/ui and mcp/packages/plugin,
since corepack only updates existing fields.
Document the canonical update procedure in
.serena/memories/workflow/updating-pnpm.md.
AI-assisted-by: omen-alpha
Two changes to the implement-plan flow:
- It now shows the whole picture before touching the repo: which
issue and branch will be created (or the current branch it
continues on), the execution style, and the task checklist — then
waits for the user's confirmation. Flow detection is read-only;
issue and branch creation moved to the execution step.
- New opt-in execution mode: on request ("step by step"), it runs
one task at a time, commits each one with the create-commit skill,
and waits for the user's review before the next task. The default
stays as it was: every task, one commit.
README and the implement-plan command description updated to match.
AI-assisted-by: omen-alpha
.agents is the shared home that opencode, Claude Code (through the
.claude/skills symlink) and Codex all read, so the skills and the
saved plans now live there instead of .opencode:
- .opencode/skills moved to .agents/skills (24 files, no content
changes).
- .opencode/plans moved to .agents/plans; the .gitignore entry
follows, so plans stay untracked.
- .claude/skills symlink retargeted to ../.agents/skills.
- planner, make-a-plan and review-plan updated to the new plans path;
new .agents/README.md documents every skill with when-to-use
examples and links to each SKILL.md.
- workflow/creating-issues memory: create-issue path updated.
opencode discovers .agents/skills natively, so .opencode needs no
reciprocal link.
AI-assisted-by: omen-alpha
Every skill in .opencode/skills now carries a "When to use" section:
triggers in any phrasing plus the matching /command for the flow
skills, one-line triggers for the utility skills, an
explicit-invocation mirror for ste, and the proactive case for
resolve-git-conflicts.
code-review-criteria drops its old usage bullets ("before merging any
PR ...") for the same role pattern as plan-review-criteria: loaded by
the reviewer subagent of the review-code flow, redirect there to
review code. Flow and criteria skills no longer compete for the same
trigger.
AI-assisted-by: omen-alpha
CLAUDE.md points Claude Code to AGENTS.md as the canonical project
instruction file, and .claude/skills symlinks the opencode skills so
both tools discover the same single source of truth. The root
.gitignore keeps ignoring new files under those paths; the tracked
entries are unaffected from now on.
AI-assisted-by: omen-alpha
Every command in .opencode/commands now only switches to the build
agent, injects the user context, and delegates to a same-named skill
that holds the full flow logic — skills are the primary mechanism for
Claude Code and Codex compatibility.
- make-a-plan, implement-plan, resolve-git-conflicts: flow logic
moved to same-named skills.
- review-plan / review-code: orchestration skills delegating to the
renamed criteria skills plan-review-criteria and
code-review-criteria, with redirecting descriptions so there is no
trigger overlap between flow and criteria.
- create-pr: absorbed the open-pr flow as Open mode (validated
branch-to-PR pipeline using scripts/detect-target-branch) plus a
new Update mode (gh pr edit); /open-pr removed.
AI-assisted-by: omen-alpha
Adds the hard rule to AGENTS.md and to mem:critical-info: CHANGES.md
is generated from GitHub milestones during the release process; it
must be updated only via the update-changelog skill flow or on
explicit user request.
AI-assisted-by: omen-alpha
/implement-plan now detects the flow from the current branch instead
of always creating an issue: on a base branch it starts standalone
(issue + branch issue-NNNN from HEAD); on a feature branch it
continues on it with no new scaffolding. Arguments override
detection (standalone, continue, no issue, from origin/<base>);
"no issue" on a base branch creates a plan-<slug> branch. Execution
is direct and the closing suggests /review-code or /open-pr.
/open-pr opens the PR for the current task branch: it detects the
base with scripts/detect-target-branch (canonical: develop, staging,
main), validates commits, issue and remote state, and stops with one
message listing everything missing. It never pushes.
AI-assisted-by: omen-alpha
Add the planning/review command suite for opencode and align the
planner skill with it:
- /make-a-plan (build agent): researches the session subject, drafts
the plan with the planner skill, asks open questions, then saves
the final plan to .opencode/plans/.
- /review-plan: reviews a plan with the plan-review skill; on
approval suggests /implement-plan, on request changes /make-a-plan.
- /review-code: reviews a diff, PR, or code change with the
code-review skill; suggests /make-a-plan for substantive findings.
All commands run as the build agent with explicit read-only rules
that expire when the command's work is done, so follow-up requests
need no agent switching. The planner skill no longer writes the plan
file on its own: it saves only when a command or the user explicitly
instructs it, and it suggests /review-plan and /implement-plan as
next steps.
AI-assisted-by: omen-alpha
* 🐛 Add content:write permission checks to Design Tokens plugin API
The Design Tokens API (tokens.cljs) had zero permission checks, allowing
any plugin to create, modify, and delete tokens, sets, and themes
regardless of granted permissions.
Add r/check-permission checks to all 22 write operations across:
- token-proxy: name, value, description, duplicate, remove, applyToken
- token-set-proxy: name, active, toggleActive, addToken, duplicate, remove
- token-theme-proxy: group, name, active, toggleActive, addSet, removeSet,
duplicate, remove
- tokens-catalog: addTheme, addSet
Follows the established pattern from comments.cljs, file.cljs, page.cljs.
Closes#11137
AI-assisted-by: qwen3.7-plus
* 🐛 Add permission checks to shape proxy interactions, detach, export, and variants
The shape proxy (shape.cljs) had multiple operations missing permission
checks, plus a cond ordering bug that bypassed the existing content:write
check for text shapes in commit-fills!.
Fix commit-fills! cond ordering: move permission check before the
text-shape branch so text shapes are also protected.
Add content:write permission checks to:
- interaction-proxy: :trigger, :delay, :action setters, :remove method
- shape-proxy: :addInteraction, :removeInteraction, :detach
- shape-proxy: :applyToken, :switchVariant, :combineAsVariants
Add content:read permission check to:
- shape-proxy: :export (read/extraction operation)
Follows the established pattern from :resize, :rotate, :blocked setters.
Relates to #11137
AI-assisted-by: qwen3.7-plus
* 🐛 Add library:write permission checks to variant plugin API
The library.cljs variant operations (variant-proxy and
lib-component-proxy) had seven mutating operations that
did not check the library:write permission, allowing
any plugin to create, modify, and delete component
variants regardless of granted permissions.
Add r/check-permission checks to all 7 operations:
- variant-proxy: addVariant, addProperty,
removeProperty, renameProperty
- lib-component-proxy: transformInVariant, addVariant,
setVariantProperty
Follows the established pattern from the :name and
:path setters in the same file.
Relates to #11137
AI-assisted-by: qwen3.7-plus
* 🐛 Add content:write permission checks to flow and flex layout plugin API
Add permission checks to prototype flow and flex layout operations
that were missing them, allowing plugins to modify flows and layout
structure without explicit user permission.
Changes:
- page.cljs: Add content:write checks to flow-proxy (name,
startingBoard setters, remove) and page-proxy (createFlow,
removeFlow)
- flex.cljs: Add content:write checks to flex-layout-proxy
(remove, appendChild)
Follows the established pattern from tokens.cljs, shape.cljs,
and library.cljs.
Relates to #11137
AI-assisted-by: qwen3.7-plus
* 🐛 Add user:read permission checks to plugin API
Add permission checks to user identity accessors that were bypassing
the consent model, allowing plugins to access user data regardless
of whether the user granted user:read permission.
Changes:
- api.cljs: Add user:read checks to getCurrentUser and getActiveUsers
- comments.cljs: Add user:read checks to comment-proxy and
comment-thread-proxy owner/user getters
- file.cljs: Add user:read check to file-version-proxy createdBy getter
When user:read permission is not granted:
- getCurrentUser() returns null
- getActiveUsers() returns empty array
- owner/user/createdBy getters return null
Follows the established pattern from other permission checks in the
plugin API.
Relates to #11137
AI-assisted-by: qwen3.7-plus
* 🐛 Fix problem with token API
---------
Co-authored-by: alonso.torres <alonso.torres@kaleidos.net>
* 🐛 Validate plugin UI URLs cannot target Penpot's own domain
The plugin UI iframe combines allow-scripts and allow-same-origin in
its sandbox. While necessary for plugins to use their own cookies and
storage, this creates a risk if a plugin's UI URL resolves to Penpot's
own origin, potentially allowing the iframe to escape sandbox isolation.
Add validateUIUrl() that checks the resolved URL against Penpot's
origin (from penpotPublicURI or location.origin) and throws if they
match. Called in openModal() after prepareUrl() resolves the URL.
Closes#11271
AI-assisted-by: qwen3.7-plus
* 💄 Fix prettier formatting in plugin-manager.spec.ts
Apply prettier formatting to fix format:check failure.
AI-assisted-by: qwen3.7-plus
* 🐛 Fix problem with penpot origin plugins
---------
Co-authored-by: alonso.torres <alonso.torres@kaleidos.net>
Prevent developer tools from being exposed when the MCP server runs in multi-user mode. Keep them available for local devenv usage and document the mode restriction. Add regression coverage for the registration policy.
Closes#11291
AI-assisted-by: gpt-5.6-luna
Co-authored-by: niwinz <843689+niwinz@users.noreply.github.com>
* 🐛 Fix font-family sample not showing for numeric font names
Setting style.fontFamily to a raw, unquoted family name (e.g. "Micro
5") parses it against CSS's <family-name> grammar: a whitespace-
separated sequence of <custom-ident>s. "Micro" tokenizes fine, but a
bare "5" isn't a valid CSS identifier (idents can't start with a
digit) — it tokenizes as a number instead, so the whole property is
invalid CSS and the browser silently drops it. Every other font in
the list happened to avoid this because none of their names have a
token that's purely numeric.
Quote the family name, matching what font-item-preview* (the font
selector's own preview, a few lines down in the same file) already
does, so it's parsed as a CSS string instead of unquoted identifiers.
Also falls back to the live fontsdb entry's family when the
typography record's own :font-family is blank — a font that was
unloaded when a typography's font/variant was last changed can leave
that field nil (the same failure mode remove-nil-style-attrs already
repairs for shape text spans) — and loads the font unconditionally in
the collapsed asset row, matching the expanded editor, since the
optical-offset cache can otherwise skip loading it entirely.
AI-assisted-by: claude-sonnet-5
* 🐛 Fix flaky typography sample position in automated tests
The optical-centering offset for the "Ag" sample (and the font
selector's fallback name label) resolves asynchronously: first paint
is unshifted, then an idle-scheduled Canvas measurement lands and the
sample jumps to its final position. Any test that checks position or
takes a screenshot shortly after paint races that jump — whether it
runs before or after is a timing accident, not a deterministic
outcome, which is exactly the "sometimes a few pixels up, sometimes
down" flakiness QA hit.
use-optical-offset now returns [offset ready?], with both lazily
initialized from the cache so a cache hit needs no async round-trip
at all. sample-text-style hides the glyphs until ready?, so the
sample only ever appears already in its final, correct position
instead of visibly moving there after the fact.
The font selector's own name label uses the same hook but always
shows real text content rather than a decorative sample, so it keeps
the old behavior instead: hiding it would blank out font names while
scrolling, worse than the minor positional nicety it's fixing.
AI-assisted-by: claude-sonnet-5
Add optional skip-onboarding param to create-demo-profile. When true, the demo profile is created with onboarding-viewed and release-notes-viewed set, so it skips the onboarding flow. Default keeps the current behavior. Cover both cases with RPC tests. AI-assisted-by: muse-spark-1.3-contributor
The creating-prs memory described a branch format (issue/... with a
<type>/<short-description> fallback) that does not match actual repo
practice, where issue-driven branches are issue-NNNN. Replace it with
a Branch Naming section: issue-NNNN as primary, descriptive name
without slashes as fallback.
Add the matching public convention to CONTRIBUTING.md under Pull
Requests, which previously defined no branch naming at all.
AI-assisted-by: omen-alpha
* 🐛 Fix crash when a nil shape id reaches WASM modifiers
A modif-tree containing a nil shape id crashed the WASM heap write
with "Cannot read properties of null (reading '__u32_buffer')" in
propagate-modifiers. Production builds elide the upstream asserts,
so a nil id (e.g. update-dimensions called with a missing shape's
parent id in the variants flow) reached the uuid heap write.
Drop nil-keyed modif-tree entries in apply-wasm-modifiers and
set-wasm-modifiers before they reach propagate-modifiers and
set-structure-modifiers, filter nil ids at the parse level, and
guard the variant reposition/creation events against missing shapes.
AI-assisted-by: hy3
* 🐛 Fix let scoping bugs in modifiers.cljs
Two let forms in set-wasm-modifiers and apply-wasm-modifiers
had their bindings vectors closed prematurely, causing undeclared
var warnings for snap-pixel?, translation?, ids, update-shape,
options, bool-ids, undo-id, and other bindings.
In set-wasm-modifiers: removed premature ) after ] on the let
bindings so the let body wraps snap-pixel? and translation?.
In apply-wasm-modifiers: the cond form was not explicitly closed,
so all subsequent bindings (ignore-tree, options, modif-tree, ids,
update-shape, bool-ids, undo-id) were parsed as cond clauses
instead of let bindings. Added ) to close cond after :else, and
moved ] to close the bindings vector after undo-id.
AI-assisted-by: mimo-v2.5-pro
* 🐛 Fix stroke width token dropdown closing mid-interaction
The stroke-row* key included a hash of applied-tokens, forcing a full
remount whenever the async token-resolution pipeline updated that
state (e.g. after a detach action settles). If the remount landed
while the width dropdown was open, it destroyed the dropdown and
reset its open state, permanently closing it before the user could
pick a token.
Drop the hash from the key so the row only remounts on actual
structural changes (add/remove/reorder), not on every token
resolution tick.
AI-assisted-by: claude-sonnet-5
* 🐛 Fix crash when a typography token has a nil value
A composite typography token saved with no fields filled in stores a
nil :value. That value reached StyleDictionary's tokens-studio
font-styles preprocessor, which assumes a typography value is never
null and threw, crashing token resolution for every other token in
the file.
Reject the nil value at the source in the typography token form
validation, and as defense in depth, filter nil-valued tokens out of
every resolve-tokens* entry point before they reach StyleDictionary,
tagging them with the existing empty-input error instead.
Also remove a stray debug prn left in find-refs.
AI-assisted-by: claude-sonnet-5
* 🔧 Gitignore local CLAUDE.md
Keep the commit-conventions instructions file local to each
contributor's checkout instead of tracking it in the repo.
AI-assisted-by: claude-sonnet-5
* ⚡ Use a single transducer to tag invalid-value tokens
merge-invalid-value-tokens ran three separate passes over the token
map (remove, map, into) and then merged the result back in. Combine
the remove/map steps into one ns-level transducer, defined once
instead of rebuilt on every call, and pass resolved as the seed to
into so the trailing merge isn't needed either.
AI-assisted-by: claude-sonnet-5
* ♻️ Drop redundant t/testing wrapper in nil-value token test
The outer t/testing just repeated the deftest's own name and added
nothing the two inner t/testing blocks (each covering one concrete
assertion group) don't already say.
AI-assisted-by: claude-sonnet-5
* 🐛 Fail the nil-value token test on a resolution error
rx/sub! only handles the success case, so if token resolution ever
errors instead, done is never called and the async test hangs
instead of failing. Switch to rx/subs! with an error handler that
reports the failure and calls done, matching the pattern already
used elsewhere in the tokens test suite.
AI-assisted-by: claude-sonnet-5
Replaces the content-hash build key (bundle_version + docker/images tree hash) used to tag and dedupe the backend/frontend/exporter/storybook/mcp image set with sha-<commit>, matching the scheme already used by admin-console, licenses-manager and payments across the org. The check→build→promote pattern with the S3 marker is unchanged; only the key used for the marker, the immutable tag and the local bundle cache filename moves from the composite build key to the git commit sha (the bundle cache now keys on bundle_version alone, which is what it actually caches). devenv is intentionally left out of this pass, it has no versioned tagging today.
Signed-off-by: David Barragán Merino <david.barragan@kaleidos.net>
The FilterInputStream proxy only implemented read() and
read(byte[], int, int). Buffered reads call read(byte[]) (and
read(byte[], int) via Clojure interop), causing ArityException
while hashing storage objects and breaking v3 imports.
Implement all read overloads and extract shared byte-count logic.
* 🐛 Fix font preview in assets breaks the font row
* 🐛 Fix font height problem also in the font dropdown
* 🐛 Fix a small bug within the changes
---------
Co-authored-by: Eva Marco <evamarcod@gmail.com>
* ⬆️ Update pnpm and its deps
* ⬆️ Update JVM dependencies in backend and common
Update several JVM dependencies across backend and common:
- passay 1.6.6 -> 2.0.0 (package reorg, ctor-based rules)
- siphash 2.0.0 -> 3.0.0 (SipHasher* renamed to SipHash*)
- lettuce-core, guava, sqlite-jdbc, jsoup, lz4-java, markdown-clj,
awssdk s3/sts, selmer, jackson-core/databind, shadow-cljs
Adapt passay validation to the new API (moved packages, constructor
configuration) and siphash to the renamed classes. Add tests for
password validation and UUID advisory-lock hashing.
AI-assisted-by: deepseek-v4-flash
* ⬆️ Update node on docker images
* 📎 Minor fixes related to pnpm12 compatibility
* 🐛 Gate MCP REPL server behind isDevEnv check
The ReplServer was starting unconditionally on every MCP server
instance, regardless of configuration. This exposed an
unauthenticated POST /execute endpoint that forwarded arbitrary
JavaScript to connected Penpot plugins.
Gate ReplServer creation, startup, and shutdown behind isDevEnv(),
consistent with how CljsReplTool and other dev tools are already
protected. Log an info message when the REPL server is disabled.
Consolidate the dev-env check into a single static isDevEnvEnabled()
method that isDevEnv() delegates to, avoiding duplicate logic. Add
PluginBridge.close() for proper WebSocket server cleanup on shutdown.
Add regression tests that construct PenpotMcpServer and verify
hasReplServer() returns the correct value based on the dev-env flag.
AI-assisted-by: mimo-v2.5-pro
* ✨ Add PENPOT_MCP_REPL_ENABLE env var for explicit REPL control
Allow the REPL server to be enabled independently of the devenv
setting via a new PENPOT_MCP_REPL_ENABLE environment variable.
When set to "true", the REPL server starts regardless of
PENPOT_MCP_DEVENV; when set to any other value, it is disabled.
When unset, the previous isDevEnv fallback applies.
Addresses review feedback on PR #11282.
AI-assisted-by: mimo-v2.5-pro
The objects-gc task was performing a hard delete on profiles
without cascading the soft-delete to owned teams, projects, and files.
This left orphaned objects that were never cleaned up.
Now the task invokes delete-object before the hard delete, ensuring
all owned resources are properly marked for deletion and cleaned up
in subsequent GC iterations.
AI-assisted-by: qwen3.7-plus
Fix two security vulnerabilities in comment RPCs when accessed
via share-links:
- GHSA-4p97-v4wg-jxfx: Share-link holders with who-comment=team
could bypass the restriction and comment. The check-comment-permissions!
function treated can-read as sufficient, but share-links always set
can-read=true.
- GHSA-fwm4-hm9f-rmcp: Comment query RPCs returned threads from all
pages, ignoring the share-link's :pages restriction.
Changes:
- files.clj: Differentiate :membership vs :share-link in
check-comment-permissions!. For share-links, require
has-comment-permissions? only (who-comment=all).
- comments.clj: Filter threads by (:pages perms) for share-link
access in get-comment-threads, get-comment-thread, and get-comments.
Closes#11370
AI-assisted-by: qwen3.7-plus
Add owner protection to ::delete-team-member RPC command.
Previously, a team admin could remove the team owner, permanently
locking them out of their team and all resources.
Changes:
- Fetch target member data before deletion
- Validate member exists (return :not-found if not)
- Reject removal if target is owner and caller is not owner
This mirrors the existing protection in update-team-member-role.
Closes#11367
AI-assisted-by: qwen3.7-plus
* 🐛 Block IPv6 transition addresses in SSRF guard
The outbound HTTP SSRF blocklist did not classify NAT64
(64:ff9b::/96), 6to4 (2002::/16) or Teredo (2001:0000::/32)
addresses, whose embedded IPv4 target is invisible to the JVM
InetAddress predicates, so URLs resolving to them could reach cloud
metadata, loopback or RFC 1918 hosts from webhook delivery and media
import.
Transition ranges are now rejected outright and any embedded IPv4 is
re-checked against the full blocklist, including operator-supplied
extra blocked CIDRs.
Closes#11319
* ♻️ Remove dead embedded-IPv4 re-check from SSRF guard
The previous commit added a recursive re-check of the IPv4 embedded in
NAT64/6to4/Teredo addresses, but the `or` in `blocked-address?`
short-circuits on the truthy keyword returned by `transition-prefix`,
so the embedded-IPv4 branch was unreachable. The transition ranges are
already rejected outright (fail-closed), making the re-check both
unnecessary and untested.
Remove `transition-embedded-ipv4`, simplify the IPv6 branch to a plain
prefix check, and correct the docstrings and tests to match what the
code actually does.
AI-assisted-by: glm-5.3-flash
* 🐛 Filter share-link tokens in get-view-only-bundle response
The get-view-only-bundle RPC command returned all share-link tokens for a file, allowing an anonymous holder of a restrictive share-link to enumerate and use more permissive tokens.
When authenticating via a share-link, the response now only includes the share-link used for authentication, preventing token disclosure and scope escalation.
Implemented using TDD:
- RED: Test demonstrates vulnerability (all tokens visible)
- GREEN: Filter share-links when (:type perms) = :share-link
- Verified all existing tests still pass
Closes#11285
AI-assisted-by: qwen3.7-plus
* 🐛 Add membership-side test for share-link token visibility
Add test coverage for the allow side of the share-link token filtering:
team members and file owners should still see all share-links, while
anonymous share-link holders only see their own token.
This protects the (:type perms) = :share-link guard from accidental
regression that could break the owner's share-link management dialog.
AI-assisted-by: qwen3.7-plus
