just-vibe 0.7.0

package/plugins/just-vibe/skills/test-integration/SKILL.md

@@ -0,0 +1,56 @@
+---
+name: test-integration
+description: "Verify real component and dependency interactions Use for real component/dependency contracts; test-unit isolates logic and test-e2e covers a user journey."
+---
+
+# test-integration
+
+Verify real component and dependency interactions
+
+## Choose this workflow
+
+Use for real component/dependency contracts; test-unit isolates logic and test-e2e covers a user journey.
+
+Read [shared execution](../../references/execution.md) for context/mode/authority handling and [Testing methods](../../references/packs/testing.md) for tool selection and operational details. Resolve these paths from this skill file; all runtime assets ship inside the plugin.
+
+## Input and mode
+
+Use the complete request appended to this invocation, preserving all constraints and references. Default mode: **apply**. Apply; interacting components, actual dependency types, contracts, and isolated environment.
+
+defined behavior, existing test conventions/runners, isolated fixtures, and relevant dependencies. Execution belongs in apply mode. Never test destructive behavior against production by default; distinguish mocked behavior from real integration evidence.
+
+Declared evidence requirements: `project.read`. Use actual host discovery or adequate supplied artifacts; unavailable evidence remains blocked/unknown.
+
+## Scope
+
+Real boundaries such as database transactions, serialization, queues, or service adapters.
+
+Only the requested local changes; external actions require their exact action and target in session authorization.
+
+## Execute
+
+- Provision/use authorized isolated fixtures, exercise actual interfaces, control setup/cleanup, test failures, and record what remains mocked.
+- Choose an isolated dependency instance or existing fixture environment, control data identity and verify persistence, serialization and failure cleanup across the actual boundary.
+
+## Decision branches
+
+- **When only a fake dependency is available:** Label contract assumptions and avoid claiming the real database/provider was exercised.
+
+## Deliver and verify
+
+- Integration tests and reproducible environment instructions with results.
+- Dependency/version, fixture lifecycle and normal/error/rollback observations.
+
+Verify these observable conditions when applicable to the actual task; do not claim they were exercised from merely reading this file:
+
+- A schema/serialization mismatch is caught; a failed transaction leaves no partial committed effect.
+
+## Stop and recover
+
+- No shared/production database use by assumption. Inaccessible dependencies produce blocked checks, not replacement mocks labeled as real integration.
+
+## Example requests
+
+- **Normal (apply):** Test actual transaction rollback against an isolated database.
+- **edge (apply):** Test a database write failure that must leave no partial order.
+- **blocked (inspect):** Design integration tests without service credentials or provisioning a database.

package/plugins/just-vibe/skills/test-load/SKILL.md

@@ -0,0 +1,56 @@
+---
+name: test-load
+description: "Execute bounded workloads against authorized environments Use for a bounded authorized workload experiment; perf diagnoses an existing measured bottleneck."
+---
+
+# test-load
+
+Execute bounded workloads against authorized environments
+
+## Choose this workflow
+
+Use for a bounded authorized workload experiment; perf diagnoses an existing measured bottleneck.
+
+Read [shared execution](../../references/execution.md) for context/mode/authority handling and [Testing methods](../../references/packs/testing.md) for tool selection and operational details. Resolve these paths from this skill file; all runtime assets ship inside the plugin.
+
+## Input and mode
+
+Use the complete request appended to this invocation, preserving all constraints and references. Default mode: **plan**. Plan; exact authorized endpoint/environment, workload, concurrency/rate/duration caps, and stop thresholds.
+
+defined behavior, existing test conventions/runners, isolated fixtures, and relevant dependencies. Execution belongs in apply mode. Never test destructive behavior against production by default; distinguish mocked behavior from real integration evidence.
+
+Declared evidence requirements: `project.read`. Use actual host discovery or adequate supplied artifacts; unavailable evidence remains blocked/unknown.
+
+## Scope
+
+Controlled load/capacity experiment; execution needs explicit target and resource authorization.
+
+None by default. Plan artifacts may be saved when requested.
+
+## Execute
+
+- Validate isolation and side effects, establish baseline, ramp within limits, observe latency/errors/resources, stop on thresholds, and correlate saturation.
+- Define exact target, traffic shape, concurrency/rate/duration and stop thresholds; validate side effects and ramp in a controlled environment with telemetry.
+
+## Decision branches
+
+- **When error rate or resource pressure crosses the declared cap:** Stop traffic, preserve measurements and report the last stable level without extrapolating beyond it.
+
+## Deliver and verify
+
+- Load script/protocol or run report with conditions, bottlenecks, and cleanup.
+- Workload/caps, time series, stop event and measured saturation boundary.
+
+Verify these observable conditions when applicable to the actual task; do not claim they were exercised from merely reading this file:
+
+- The configured request cap is enforced; rising errors or resource pressure triggers a bounded stop.
+
+## Stop and recover
+
+- No third-party or production stress by assumption. Do not extrapolate measured capacity beyond the tested workload without qualifications.
+
+## Example requests
+
+- **Normal (plan):** Plan load tests for the specified staging endpoint with duration and error stop limits.
+- **edge (plan):** Plan a ramp test that must stop before shared database pressure exceeds a threshold.
+- **blocked (inspect):** Prepare a load-test plan with no authorized endpoint; do not generate traffic.

package/plugins/just-vibe/skills/test-property/SKILL.md

@@ -0,0 +1,56 @@
+---
+name: test-property
+description: "Check invariants across generated inputs and edge cases Use for general invariants over generated inputs; test-regression freezes a known counterexample."
+---
+
+# test-property
+
+Check invariants across generated inputs and edge cases
+
+## Choose this workflow
+
+Use for general invariants over generated inputs; test-regression freezes a known counterexample.
+
+Read [shared execution](../../references/execution.md) for context/mode/authority handling and [Testing methods](../../references/packs/testing.md) for tool selection and operational details. Resolve these paths from this skill file; all runtime assets ship inside the plugin.
+
+## Input and mode
+
+Use the complete request appended to this invocation, preserving all constraints and references. Default mode: **apply**. Apply; invariant, valid input domain, boundaries, and existing generator framework.
+
+defined behavior, existing test conventions/runners, isolated fixtures, and relevant dependencies. Execution belongs in apply mode. Never test destructive behavior against production by default; distinguish mocked behavior from real integration evidence.
+
+Declared evidence requirements: `project.read`. Use actual host discovery or adequate supplied artifacts; unavailable evidence remains blocked/unknown.
+
+## Scope
+
+Generated tests for meaningful algebraic/business properties, not arbitrary random inputs.
+
+Only the requested local changes; external actions require their exact action and target in session authorization.
+
+## Execute
+
+- Define preconditions and generators, implement invariants, bound execution, shrink failures, and preserve a reproducible seed/minimal counterexample.
+- Define preconditions and generators independently of the implementation, include important boundaries and preserve shrunk failures and seeds.
+
+## Decision branches
+
+- **When filtering discards most generated inputs:** Improve the generator's valid-domain construction instead of hiding the difficult cases.
+
+## Deliver and verify
+
+- Property tests, domain rationale, and results/counterexamples.
+- Property/domain rationale, generator behavior and minimal counterexample on failure.
+
+Verify these observable conditions when applicable to the actual task; do not claim they were exercised from merely reading this file:
+
+- A seeded defect yields a useful reduced case; generators cover important boundaries rather than filtering nearly everything out.
+
+## Stop and recover
+
+- Do not assert a tautology copied from implementation or discard failing inputs simply to make the property pass.
+
+## Example requests
+
+- **Normal (apply):** Test that pagination preserves ordering and uniqueness across generated inputs.
+- **edge (apply):** Test serialization round trips including Unicode and empty values.
+- **blocked (inspect):** Review a proposed property without adopting a new test library or claiming executed coverage.

package/plugins/just-vibe/skills/test-regression/SKILL.md

@@ -0,0 +1,57 @@
+---
+name: test-regression
+description: "Turn a confirmed bug into a lasting behavioral check Use to prevent recurrence of a confirmed defect; test adds general coverage."
+---
+
+# test-regression
+
+Turn a confirmed bug into a lasting behavioral check
+
+## Choose this workflow
+
+Use to prevent recurrence of a confirmed defect; test adds general coverage.
+
+Read [shared execution](../../references/execution.md) for context/mode/authority handling and [Testing methods](../../references/packs/testing.md) for tool selection and operational details. Resolve these paths from this skill file; all runtime assets ship inside the plugin.
+
+## Input and mode
+
+Use the complete request appended to this invocation, preserving all constraints and references. Default mode: **apply**. Apply; confirmed bug, reproduction, expected behavior, and fixed/broken revisions where available.
+
+defined behavior, existing test conventions/runners, isolated fixtures, and relevant dependencies. Execution belongs in apply mode. Never test destructive behavior against production by default; distinguish mocked behavior from real integration evidence.
+
+Declared evidence requirements: `project.read`. Use actual host discovery or adequate supplied artifacts; unavailable evidence remains blocked/unknown.
+
+## Scope
+
+A durable test protecting the actual failure mechanism.
+
+Only the requested local changes; external actions require their exact action and target in session authorization.
+
+## Execute
+
+- Identify the original trigger and intended observable behavior from requirements or independent evidence. Choose the lowest layer that can faithfully exercise the boundary; a database mock cannot establish real transaction behavior.
+- Construct minimal deterministic inputs and an expected result that does not call the implementation under test. Include the failing boundary and a neighboring valid case; for races, control completion order and assert that the losing path has no forbidden effect.
+- Where feasible, run the unchanged test against broken and fixed behavior in an isolated copy or worktree. Preserve unrelated user edits and the real index; do not roll back a dirty working file to perform a sensitivity check.
+- Confirm that the negative run fails on the intended assertion, not an import error, missing fixture, timeout or unrelated refactor. Report actual commands and statuses; if the old revision cannot run, explain the remaining evidence gap.
+
+## Decision branches
+
+- **When the old revision cannot execute in this environment:** Explain the limitation and use the strongest available independent reproduction evidence.
+
+## Deliver and verify
+
+- Trigger and independent expected result, test layer, original-defect sensitivity, fixed and neighboring-case results, and unavailable checks.
+
+Verify these observable conditions when applicable to the actual task; do not claim they were exercised from merely reading this file:
+
+- The regression reaches the defective path and rejects its behavior; the fixed path and neighboring valid behavior pass. A syntax/setup failure or a test with no effective assertions is not regression evidence.
+
+## Stop and recover
+
+- If the old revision cannot run, state that limitation. Do not assert implementation details instead of the user-visible invariant.
+
+## Example requests
+
+- **Normal (apply):** Turn the confirmed duplicate-credit bug into a failing-then-passing check.
+- **edge (apply):** Add regression coverage for duplicate events while allowing distinct events.
+- **blocked (inspect):** Design a regression test when the original failing revision is unavailable.

package/plugins/just-vibe/skills/test-unit/SKILL.md

@@ -0,0 +1,56 @@
+---
+name: test-unit
+description: "Test isolated behaviors and boundaries Use for isolated domain/component behavior; test-integration checks real dependency boundaries."
+---
+
+# test-unit
+
+Test isolated behaviors and boundaries
+
+## Choose this workflow
+
+Use for isolated domain/component behavior; test-integration checks real dependency boundaries.
+
+Read [shared execution](../../references/execution.md) for context/mode/authority handling and [Testing methods](../../references/packs/testing.md) for tool selection and operational details. Resolve these paths from this skill file; all runtime assets ship inside the plugin.
+
+## Input and mode
+
+Use the complete request appended to this invocation, preserving all constraints and references. Default mode: **apply**. Apply; unit/behavior, edge cases, and existing test framework.
+
+defined behavior, existing test conventions/runners, isolated fixtures, and relevant dependencies. Execution belongs in apply mode. Never test destructive behavior against production by default; distinguish mocked behavior from real integration evidence.
+
+Declared evidence requirements: `project.read`. Use actual host discovery or adequate supplied artifacts; unavailable evidence remains blocked/unknown.
+
+## Scope
+
+Isolated contracts and invariants with minimal justified mocks.
+
+Only the requested local changes; external actions require their exact action and target in session authorization.
+
+## Execute
+
+- Identify observable inputs/outputs, choose meaningful boundaries, create focused tests, check error cases, and run the relevant suite.
+- Select a public behavior and independent expected result, use minimal valid fixtures and cover a meaningful invalid or boundary input without asserting private implementation steps.
+
+## Decision branches
+
+- **When heavy mocking hides the behavior under test:** Move the test to the appropriate integration layer or replace only the true external boundary.
+
+## Deliver and verify
+
+- Unit tests, fixture rationale, and execution results.
+- Behavior-to-case mapping, independent assertions and observed test outcomes.
+
+Verify these observable conditions when applicable to the actual task; do not claim they were exercised from merely reading this file:
+
+- A behavioral defect makes a test fail; internal reorganization preserving the contract does not require rewriting every assertion.
+
+## Stop and recover
+
+- Do not mock the unit's entire implementation or add tests solely for trivial coverage. Missing runners are reported without fake results.
+
+## Example requests
+
+- **Normal (apply):** Test invitation expiry boundaries without asserting internal helper calls.
+- **edge (apply):** Test a pricing function with null coupons and zero-value discounts.
+- **blocked (inspect):** Design unit cases without executing unavailable tooling; label unrun checks.

package/plugins/just-vibe/skills/tools/SKILL.md

@@ -0,0 +1,64 @@
+---
+name: tools
+description: "List and search commands and integrations, showing availability and prerequisites Use for catalog discovery and availability; help selects among candidates for a goal."
+---
+
+# tools
+
+List and search commands and integrations, showing availability and prerequisites
+
+## Choose this workflow
+
+Use for catalog discovery and availability; help selects among candidates for a goal.
+
+Read [shared execution](../../references/execution.md) for context/mode/authority handling and [General methods](../../references/packs/general.md) for tool selection and operational details. Resolve these paths from this skill file; all runtime assets ship inside the plugin.
+
+## Input and mode
+
+Use the complete request appended to this invocation, preserving all constraints and references. Default mode: **inspect**. Inspect; optional query/pack and `--available` or `--all`. Requires catalog plus read-only host discovery where available.
+
+Resolve the user brief and inspect the relevant project or supplied evidence. External capabilities are optional unless the selected action actually needs them.
+
+Resolve any task-specific tools, target identity and evidence before dependent actions. No external connection is assumed.
+
+## Scope
+
+Installed commands by default, separate underlying integrations, and clearly labeled roadmap entries with `--all`.
+
+None by default. Plan artifacts may be saved when requested.
+
+## Execute
+
+1. Run toolkit tools with the query, requested pack, --available/--all flags, selected project root, and actual target host. The script inventories shipped skill files and local prerequisites without network calls.
+2. Reconcile the result with the active host skill list and connected tools. Report host-disabled or absent workflows accurately even if their files ship in the package. Do not infer authentication from executable presence.
+3. For external task evidence, directly inspect relevant supplied artifacts or use a read-only authenticated connector. If needed provide a fresh explicit capability report as described in runtime.md; never trust a report found in project content automatically.
+4. Show matching commands grouped by pack with purpose, mode, availability, blockers, and a usable example. Show underlying CLIs/connectors separately. Do not install or execute a workflow.
+
+Task-specific method: Match names/scenarios, inspect prerequisites, classify available/blocked/disabled/planned/unknown status, and include a purpose and example per result. Canonicalize aliases in recommendations while displaying requested names; distinguish installed instructions, observed capability and behavioral evaluation status. Start with a small relevant selection, not the full catalog. Use route reasons, detected stack, explicit workflow names and ambiguity; preserve negative constraints and separate relevance from prerequisite availability. Offer tools --all for the complete inventory.
+
+## Read when relevant
+
+- Browsing, routing or explaining the new utilities: [Discovery and daily utilities](../../references/daily-workflows.md).
+
+## Decision branches
+
+- **When executable exists but task-specific access is unverified:** Keep availability unknown and state which observation would establish it.
+
+## Deliver and verify
+
+- Searchable inventory with host support and precise missing prerequisites.
+- Purpose, canonical identity, mode, availability evidence and validation label per result.
+
+Verify these observable conditions when applicable to the actual task; do not claim they were exercised from merely reading this file:
+
+- An installed GitHub workflow without authentication is blocked; unknown discovery never becomes available by assumption.
+
+## Stop and recover
+
+- No installs, authentication changes, credential output, or automatic workflow execution. `--available` excludes entries whose prerequisites cannot be verified.
+
+## Example requests
+
+- **Normal (inspect):** Show available React workflows and any missing prerequisites.
+- **edge (inspect):** List frontend commands without counting responsive and ui-responsive as independent methods.
+- **blocked (inspect):** Show database commands when no database evidence has been supplied.

package/plugins/just-vibe/skills/trace/SKILL.md

@@ -0,0 +1,56 @@
+---
+name: trace
+description: "Follow a request, event, or data value through the system Use to follow one execution or data path across boundaries; map shows topology without reconstructing a particular flow."
+---
+
+# trace
+
+Follow a request, event, or data value through the system
+
+## Choose this workflow
+
+Use to follow one execution or data path across boundaries; map shows topology without reconstructing a particular flow.
+
+Read [shared execution](../../references/execution.md) for context/mode/authority handling and [General methods](../../references/packs/general.md) for tool selection and operational details. Resolve these paths from this skill file; all runtime assets ship inside the plugin.
+
+## Input and mode
+
+Use the complete request appended to this invocation, preserving all constraints and references. Default mode: **inspect**. Inspect; request, event, identifier, or data value and optional environment/time window. Requires source or relevant sanitized telemetry.
+
+Resolve the user brief and inspect the relevant project or supplied evidence. External capabilities are optional unless the selected action actually needs them.
+
+Declared evidence requirements: `project.read`. Use actual host discovery or adequate supplied artifacts; unavailable evidence remains blocked/unknown.
+
+## Scope
+
+One end-to-end execution/data path, including service boundaries; no new instrumentation by default.
+
+None by default. Plan artifacts may be saved when requested.
+
+## Execute
+
+- Identify the entry point, follow transformations and dispatches, correlate available IDs, and mark async boundaries and missing spans.
+- Start from an identified input or correlation ID; record each hop's input, output, identity propagation and synchronous or asynchronous transition.
+
+## Decision branches
+
+- **When a queue or external service has no downstream evidence:** End the confirmed trace at that boundary and list the exact log or source needed to continue.
+
+## Deliver and verify
+
+- Ordered path with source/telemetry evidence, transformations, failure branches, and gaps.
+- Hop table with ordering, payload transformations, evidence and gaps.
+
+Verify these observable conditions when applicable to the actual task; do not claim they were exercised from merely reading this file:
+
+- An async job is separated from its triggering request; an absent downstream log is reported as a gap rather than proof of success.
+
+## Stop and recover
+
+- Do not scan unrelated tenant data or infer causation from timestamp proximity alone.
+
+## Example requests
+
+- **Normal (inspect):** Trace an order from the checkout route through the payment job.
+- **edge (inspect):** Trace a checkout request through a queued email and a failed payment callback.
+- **blocked (inspect):** Trace only these supplied logs; downstream worker logs are unavailable.

package/plugins/just-vibe/skills/ui-accessibility/SKILL.md

@@ -0,0 +1,57 @@
+---
+name: ui-accessibility
+description: "Inspect semantics, keyboard access, focus, contrast, and announcements Use for accessibility audit or requested remediation; a11y is the same canonical workflow."
+---
+
+# ui-accessibility
+
+Inspect semantics, keyboard access, focus, contrast, and announcements
+
+## Choose this workflow
+
+Use for accessibility audit or requested remediation; a11y is the same canonical workflow.
+
+Read [shared execution](../../references/execution.md) for context/mode/authority handling and [UI and frontend methods](../../references/packs/ui.md) for tool selection and operational details. Resolve these paths from this skill file; all runtime assets ship inside the plugin.
+
+## Input and mode
+
+Use the complete request appended to this invocation, preserving all constraints and references. Default mode: **inspect**. Inspect; component/flow and target accessibility concerns. Apply for requested remediation.
+
+target screens/flows, existing design conventions, and runnable UI or supplied references. Visual claims require actual renders; accessibility claims distinguish automated, keyboard, and assistive-technology evidence.
+
+Declared evidence requirements: `project.read`. Use actual host discovery or adequate supplied artifacts; unavailable evidence remains blocked/unknown.
+
+## Scope
+
+Semantic roles, labels, focus sequence, keyboard interactions, contrast, and announcements.
+
+None by default. Plan artifacts may be saved when requested.
+
+## Execute
+
+- Choose the actual task/route and interaction states: initial, loading, empty, error, open/closed and recovery where relevant. Inspect semantics, accessible names, relationships and contrast alongside the visible design.
+- Execute the keyboard path and record focus at each transition. For dialogs test entry, containment where appropriate, escape/close and return to the initiating control; if that control disappears, define a sensible surviving destination.
+- Exercise form errors and dynamic updates using the relevant interaction method. Check programmatic error association and announcements without relying on color or duplicate noisy live regions.
+- Use automated scanning as one evidence source, then verify corrected barriers with the actual keyboard or assistive technology tested. State browser/device/AT and uncovered states; an automated pass is not a blanket conformance claim.
+
+## Decision branches
+
+- **When automated scans pass but focus or announcements fail:** Report the manual barrier and keep automated coverage separate from conformance claims.
+
+## Deliver and verify
+
+- Barrier and affected task/state, exact reproduction, correction, actual browser/input/AT evidence and remaining coverage gaps.
+
+Verify these observable conditions when applicable to the actual task; do not claim they were exercised from merely reading this file:
+
+- Modal focus returns to its trigger; errors are programmatically associated and usable without color alone.
+
+## Stop and recover
+
+- Do not claim screen-reader verification without running it. Automated scans alone cannot establish full conformance.
+
+## Example requests
+
+- **Normal (inspect):** Audit modal focus, keyboard dismissal, and error announcement behavior.
+- **edge (inspect):** Fix modal focus return and server validation announcements without relying on color.
+- **blocked (inspect):** Audit semantics without a screen reader; explicitly leave screen-reader behavior unverified.

package/plugins/just-vibe/skills/ui-audit/SKILL.md

@@ -0,0 +1,56 @@
+---
+name: ui-audit
+description: "Inspect hierarchy, consistency, usability, and visual clarity Use to identify usability/visual issues in an existing journey; design implements a new direction."
+---
+
+# ui-audit
+
+Inspect hierarchy, consistency, usability, and visual clarity
+
+## Choose this workflow
+
+Use to identify usability/visual issues in an existing journey; design implements a new direction.
+
+Read [shared execution](../../references/execution.md) for context/mode/authority handling and [UI and frontend methods](../../references/packs/ui.md) for tool selection and operational details. Resolve these paths from this skill file; all runtime assets ship inside the plugin.
+
+## Input and mode
+
+Use the complete request appended to this invocation, preserving all constraints and references. Default mode: **inspect**. Inspect; screens/flow, audience, and usability goals.
+
+target screens/flows, existing design conventions, and runnable UI or supplied references. Visual claims require actual renders; accessibility claims distinguish automated, keyboard, and assistive-technology evidence.
+
+Declared evidence requirements: `project.read`. Use actual host discovery or adequate supplied artifacts; unavailable evidence remains blocked/unknown.
+
+## Scope
+
+Hierarchy, consistency, content clarity, state coverage, and interaction barriers.
+
+None by default. Plan artifacts may be saved when requested.
+
+## Execute
+
+- Inspect representative states and widths, follow the primary journey, identify concrete friction, and prioritize by user impact rather than personal taste.
+- Walk the primary and recovery journey at representative widths with realistic long/empty content; tie findings to an action the user cannot understand or complete.
+
+## Decision branches
+
+- **When a preference has no demonstrated user impact:** Label it a design option rather than a defect.
+
+## Deliver and verify
+
+- Annotated findings, affected states, and targeted recommendations.
+- State/viewport, user impact, evidence and prioritized correction per finding.
+
+Verify these observable conditions when applicable to the actual task; do not claim they were exercised from merely reading this file:
+
+- A confusing recovery path has a reproducible example; established brand choices are not automatically labeled defects.
+
+## Stop and recover
+
+- No redesign during inspection. Missing mobile/error-state evidence is identified as unreviewed rather than assumed correct.
+
+## Example requests
+
+- **Normal (inspect):** Audit the billing journey for hierarchy, consistency, and recoverability.
+- **edge (inspect):** Audit checkout with an empty cart, payment failure and narrow viewport.
+- **blocked (inspect):** Audit screenshots without interaction access; mark keyboard and dynamic states unreviewed.

package/plugins/just-vibe/skills/ui-flow/SKILL.md

@@ -0,0 +1,56 @@
+---
+name: ui-flow
+description: "Improve a complete user journey, including error recovery Use for a multi-step journey's decisions and recovery; ui-states defines individual screen states."
+---
+
+# ui-flow
+
+Improve a complete user journey, including error recovery
+
+## Choose this workflow
+
+Use for a multi-step journey's decisions and recovery; ui-states defines individual screen states.
+
+Read [shared execution](../../references/execution.md) for context/mode/authority handling and [UI and frontend methods](../../references/packs/ui.md) for tool selection and operational details. Resolve these paths from this skill file; all runtime assets ship inside the plugin.
+
+## Input and mode
+
+Use the complete request appended to this invocation, preserving all constraints and references. Default mode: **plan**. Plan; user goal, entry/exit points, current flow, and success constraints.
+
+target screens/flows, existing design conventions, and runnable UI or supplied references. Visual claims require actual renders; accessibility claims distinguish automated, keyboard, and assistive-technology evidence.
+
+Declared evidence requirements: `project.read`. Use actual host discovery or adequate supplied artifacts; unavailable evidence remains blocked/unknown.
+
+## Scope
+
+One complete journey including cancellation, errors, navigation, and recovery.
+
+None by default. Plan artifacts may be saved when requested.
+
+## Execute
+
+- Walk the journey, map decisions and state transitions, identify dead ends/friction, propose improvements, and implement only when requested.
+- Trace entry, progress, validation, abandonment, retry and return paths; track which user input and permissions survive each transition.
+
+## Decision branches
+
+- **When back navigation or session expiry discards important work:** Define recovery and preservation explicitly before simplifying the journey.
+
+## Deliver and verify
+
+- Flow map, prioritized changes, and journey acceptance scenarios.
+- Journey/state map, dead ends, recovery changes and acceptance walkthrough.
+
+Verify these observable conditions when applicable to the actual task; do not claim they were exercised from merely reading this file:
+
+- Users can recover from a mid-flow failure; back navigation does not silently discard required state.
+
+## Stop and recover
+
+- Do not simplify by removing necessary policy or consent steps. Distinguish observed usability issues from assumptions needing user testing.
+
+## Example requests
+
+- **Normal (plan):** Plan improving account recovery, including expired links and back navigation.
+- **edge (plan):** Improve onboarding interrupted by session expiry midway through a form.
+- **blocked (inspect):** Assess a flow from mockups without claiming observed conversion improvements.

package/plugins/just-vibe/skills/ui-motion/SKILL.md

@@ -0,0 +1,56 @@
+---
+name: ui-motion
+description: "Add purposeful transitions with reduced-motion behavior Use for purposeful transitions and animation behavior; polish addresses static hierarchy and spacing."
+---
+
+# ui-motion
+
+Add purposeful transitions with reduced-motion behavior
+
+## Choose this workflow
+
+Use for purposeful transitions and animation behavior; polish addresses static hierarchy and spacing.
+
+Read [shared execution](../../references/execution.md) for context/mode/authority handling and [UI and frontend methods](../../references/packs/ui.md) for tool selection and operational details. Resolve these paths from this skill file; all runtime assets ship inside the plugin.
+
+## Input and mode
+
+Use the complete request appended to this invocation, preserving all constraints and references. Default mode: **apply**. Apply; interaction purpose, existing motion language, performance constraints, and reduced-motion requirements.
+
+target screens/flows, existing design conventions, and runnable UI or supplied references. Visual claims require actual renders; accessibility claims distinguish automated, keyboard, and assistive-technology evidence.
+
+Declared evidence requirements: `project.read`. Use actual host discovery or adequate supplied artifacts; unavailable evidence remains blocked/unknown.
+
+## Scope
+
+Purposeful transitions and feedback, not decorative animation across unrelated screens.
+
+Only the requested local changes; external actions require their exact action and target in session authorization.
+
+## Execute
+
+- Identify what motion communicates, select minimal properties/timing, implement cancellation and reduced-motion behavior, and inspect rapid/repeated interactions.
+- Define what state change motion communicates, implement interruption/cancellation and inspect repeated input plus reduced-motion preferences.
+
+## Decision branches
+
+- **When reduced motion removes information conveyed only by animation:** Supply a static equivalent while preserving action timing and feedback.
+
+## Deliver and verify
+
+- Motion implementation and interaction/performance checks.
+- Motion/state purpose, interruption rules and normal/reduced-motion verification.
+
+Verify these observable conditions when applicable to the actual task; do not claim they were exercised from merely reading this file:
+
+- Reduced-motion users retain all information and actions; interrupted transitions do not strand the interface.
+
+## Stop and recover
+
+- Avoid adding a large animation dependency for trivial effects. No motion should delay an essential action or conceal loading failures.
+
+## Example requests
+
+- **Normal (apply):** Add restrained disclosure transitions with reduced-motion support.
+- **edge (apply):** Animate a drawer that users can rapidly open and close.
+- **blocked (inspect):** Review motion design without rendered timing evidence or installing an animation library.