Repository navigation
chore(upstream): merge cursor/plugins - #32
Open
mstephenholl wants to merge 33 commits into
Open
mstephenholl wants to merge 33 commits into
mstephenholl wants to merge 33 commits into
Conversation
X tools are included in Grok Bot, so cost estimates, credit checks, and budget tiers only apply to webhooks and X Activity API subscriptions now. The guide is rewritten from scratch at about half the length, and the pricing reference is removed. Co-authored-by: Cursor <[email protected]>
Unlocking as soon as the PIN arrives meant a later Yes never saved the key. Co-authored-by: Cursor <[email protected]>
Every X tool is treated as free, so a credits error is an unexpected X error rather than a prompt to buy credits. Co-authored-by: Cursor <[email protected]>
Search points at search_posts_all with start_time, the reading table names actual tools and adds trends, missing public-read tools on the user's connection aren't an error, and unlock checks the saved key and stored PIN before asking. Co-authored-by: Cursor <[email protected]>
Rebuild the X MCP guide around free X tools
Co-authored-by: Cursor <[email protected]>
Adds the 23 new reads and list writes to the connect message, reading table, and a Lists section, and notes there's no list delete or community-posts read. Co-authored-by: Cursor <[email protected]>
feat(third_party): add Jobber plugin
Frontmatter reconnect hint names Spaces and Lists, the built-in reads paragraph says they need the user's connection, plugin.json describes them, and small wording nits. Co-authored-by: Cursor <[email protected]>
Co-authored-by: Cursor <[email protected]>
…es-lists X guide: Spaces, Communities, and Lists
[databricks-genie-one] Add Databricks Genie One plugin
…logo [databricks-genie-one] Use the Genie icon as the logo
Co-authored-by: Cursor <[email protected]>
feat(quickbooks-online): add Intuit QuickBooks skills
…s-skills Revert "feat(quickbooks-online): add Intuit QuickBooks skills"
Co-authored-by: Cursor <[email protected]>
Co-authored-by: Cursor <[email protected]>
Co-authored-by: Cursor <[email protected]>
…client-d351 google-cloud-bigquery: add team OAuth client variables
Co-authored-by: Cursor <[email protected]>
Add Lark and Feishu official remote MCP plugins
[framer] Add Framer plugin
Co-authored-by: Cursor <[email protected]>
Co-authored-by: Cursor <[email protected]>
Add Vanta plugin
…ursor#741) * Add the statechart skill to pstack Draws what a user sees as a Mermaid statechart generated from a TLA+ model. TLC checks the reader's rules as invariants. Each state is marked works, still wrong, or fixed by an open change, and each counterexample class is drawn as a numbered path. The skill also turns an agreed chart into a TLA+ skeleton and compares the drawn chart with the agreed one. Bumps pstack to 0.15.30 and regenerates the Claude Code harness tree. * Fix the review findings in the statechart skill - Check each rule in its own TLC run. The generated cfg names no rule, and a comment in it lists the rules. - Refuse TLC output from a run that failed or stopped early, and read counterexamples that a start state breaks. - Count steps that change nothing when a state is settled, number bugs that only the fixed model has after the current bugs, and escape quotes in captions. - Compare parallel regions as one machine, and refuse charts that were not drawn for comparison. - Refuse final states and duplicate captions, read class shorthand and states first named inside a composite, and match a never line to an exact state. - Rewrite SKILL.md to match, and run the skill script tests in CI. * Refuse a -- outside a composite in statechart charts A -- at the top level of a chart opened a second root part that the parser never read, so every region after it was dropped. chart-to-tla wrote a one-region skeleton, and compare-charts could print "same". parseMermaid now refuses the chart. A unit test covers parseMermaid, chart-to-tla, and compare-charts. Co-authored-by: lauren <[email protected]> * Exit 2 when compare-charts cannot read a chart compare-charts let a read or parse error escape, so a chart it could not read exited 1, the code for charts that differ. It now prints the file and the reason to stderr and exits 2, as SKILL.md says for a refused chart. Exit 1 now means only that the charts differ. Co-authored-by: lauren <[email protected]> * Update the statechart skill to its current version Replace the skill, its scripts, and its tests with the current version. statechart.mjs runs the whole flow with from-spec and from-chart, and its --help replaces the old Setup section. New features are a code-against-spec audit (--against), nested state boxes, parallel regions, guards, scoped charts, timer labels, and legend groups and colors chosen in the mapping. TLC comes from TLA2TOOLS and JAVA only. Without TLA2TOOLS, the scripts point to the TLA+ releases page and exit 2. Co-authored-by: lauren <[email protected]> * Exit 2 when --png cannot render a chart A failed mermaid-cli render printed "could not render" but still exited 0, or 0 with --expect-broken, so callers reported PNGs that did not exist. finish() now exits 2 whenever --png renders fewer PNGs than charts, and keeps the files it wrote. The --help exit codes say so, and a new test covers it with an npx that cannot launch a browser. Co-authored-by: lauren <[email protected]> * Update the statechart skill to its latest version Rendered charts keep their full width. An edge label writes the opening words that its choices share once, instead of cutting the label with "…". --full-labels replaces --all-labels, and compare-charts refuses a chart drawn without it. The fix replay matches each bug step by the screens and step words that the reader sees, and two states get at most one arrow. Each rule also runs as a TLC CONSTRAINT, so a rule that stays broken stops at its first bad state instead of flooding TLC with counterexamples. statechart.mjs passes its arguments and models through functions instead of module globals. The --png exit code from the previous commit is now part of the source. Co-authored-by: lauren <[email protected]> * Update the statechart skill: safe output folder, its TLA+ model, and the comment and assignment cleanups Co-authored-by: lauren <[email protected]> * Drop a comment in the public findTlc that only describes the code Co-authored-by: lauren <[email protected]> --------- Co-authored-by: Cursor Agent <[email protected]>
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
5 participants
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Upstream: cursor/plugins@eb4db2b55b2b
Dropped 67 upstream paths outside the port.
Needs porting
git rm -r pstack/skills/statechart.Merging records upstream progress, so the next sync only sees newer changes. Run
pstack/port/smoke/smoke.pybefore merging if a ported skill changed.