feat: vendor FIRST 0.3 and thin station overlays

.agents/skills/b/SKILL.md

@@ -0,0 +1,86 @@
+---
+name: b
+description: Basilic workflow dispatcher and catalog. Use when the user types /b or /b-<playbook>.
+disable-model-invocation: true
+---
+
+# Basilic workflows
+
+With no argument, list the shortcuts below and stop. Do not start a lifecycle or execute all playbooks.
+
+For `/b <token>`, resolve a shortcut below or a full name from the index (with or without `b-`), read that child SKILL.md, and follow it. Preserve the remaining request as its task context. Unknown token: show the index and stop; never guess a publishing command.
+
+Direct `/b-<name>` loads the same child. Names in an old conversation such as `/plan-feature` can be translated to `/b-plan-feature`; they are not separate installed aliases. FIRST `/f-*` owns durable decisions and remains a separate catalog. Install with `pnpm dlx skills@latest add blockmatic/first` ([reviewed revision](https://github.com/blockmatic/first/tree/d67d26c59c5501c4f6b8518d8543c0721e98f10d)). Do not fork that tree here. Use the repository's FIRST load order when that boundary is relevant.
+
+## Shortcuts
+
+| Invocation | Playbook |
+|---|---|
+| `/b plan` | [b-plan-feature](b-plan-feature/SKILL.md) |
+| `/b build` | [b-build](b-build/SKILL.md) |
+| `/b review` | [b-code-review](b-code-review/SKILL.md) |
+| `/b debug` | [b-debug-issue](b-debug-issue/SKILL.md) |
+| `/b test` | [b-run-all-tests-and-fix](b-run-all-tests-and-fix/SKILL.md) |
+| `/b commit` | [b-git-commit](b-git-commit/SKILL.md) |
+| `/b push` | [b-git-push](b-git-push/SKILL.md) |
+| `/b pr` | [b-git-create-pr](b-git-create-pr/SKILL.md) |
+| `/b retro` | [b-retro](b-retro/SKILL.md) |
+
+`build` ends at verified local changes. `commit`, `push`, `pr`, and `exec-push` request their named Git actions; none requests merging or deploying.
+
+## Full index
+
+- [/b-add-documentation](b-add-documentation/SKILL.md)
+- [/b-add-error-handling](b-add-error-handling/SKILL.md)
+- [/b-audit-accessibility](b-audit-accessibility/SKILL.md)
+- [/b-build](b-build/SKILL.md)
+- [/b-clarify-task](b-clarify-task/SKILL.md)
+- [/b-code-review](b-code-review/SKILL.md)
+- [/b-coderabbit](b-coderabbit/SKILL.md)
+- [/b-council](b-council/SKILL.md)
+- [/b-debug-browser](b-debug-browser/SKILL.md)
+- [/b-debug-issue](b-debug-issue/SKILL.md)
+- [/b-deslop](b-deslop/SKILL.md)
+- [/b-diagrams](b-diagrams/SKILL.md)
+- [/b-docker-logs](b-docker-logs/SKILL.md)
+- [/b-exec-push](b-exec-push/SKILL.md)
+- [/b-fix-compile-errors](b-fix-compile-errors/SKILL.md)
+- [/b-fix-git-issues](b-fix-git-issues/SKILL.md)
+- [/b-fix-github-actions](b-fix-github-actions/SKILL.md)
+- [/b-fix-push](b-fix-push/SKILL.md)
+- [/b-fix-vercel-build](b-fix-vercel-build/SKILL.md)
+- [/b-generate-api-docs](b-generate-api-docs/SKILL.md)
+- [/b-generate-pr-description](b-generate-pr-description/SKILL.md)
+- [/b-git-commit](b-git-commit/SKILL.md)
+- [/b-git-create-pr](b-git-create-pr/SKILL.md)
+- [/b-git-pr-comments](b-git-pr-comments/SKILL.md)
+- [/b-git-push](b-git-push/SKILL.md)
+- [/b-info-architecture](b-info-architecture/SKILL.md)
+- [/b-light-review-existing-diffs](b-light-review-existing-diffs/SKILL.md)
+- [/b-lint-fix](b-lint-fix/SKILL.md)
+- [/b-lint-suite](b-lint-suite/SKILL.md)
+- [/b-nextjs-form](b-nextjs-form/SKILL.md)
+- [/b-onboard-new-developer](b-onboard-new-developer/SKILL.md)
+- [/b-optimize-performance](b-optimize-performance/SKILL.md)
+- [/b-overview](b-overview/SKILL.md)
+- [/b-plan-architecture](b-plan-architecture/SKILL.md)
+- [/b-plan-feature](b-plan-feature/SKILL.md)
+- [/b-refactor-code](b-refactor-code/SKILL.md)
+- [/b-release-review](b-release-review/SKILL.md)
+- [/b-retro](b-retro/SKILL.md)
+- [/b-review-plan](b-review-plan/SKILL.md)
+- [/b-roadmap](b-roadmap/SKILL.md)
+- [/b-run-all-tests-and-fix](b-run-all-tests-and-fix/SKILL.md)
+- [/b-security-audit](b-security-audit/SKILL.md)
+- [/b-security-review](b-security-review/SKILL.md)
+- [/b-use-shadcn](b-use-shadcn/SKILL.md)
+- [/b-use-tdd](b-use-tdd/SKILL.md)
+- [/b-use-v0](b-use-v0/SKILL.md)
+- [/b-visualize](b-visualize/SKILL.md)
+- [/b-write-api-test](b-write-api-test/SKILL.md)
+- [/b-write-unit-tests](b-write-unit-tests/SKILL.md)
+- [/b-yolo](b-yolo/SKILL.md)
+
+## Authoring
+
+For skill changes, read [the authoring pattern](references/authoring.md). For delivery evidence, read [completion evidence](references/completion.md). Both ship inside this installable tree.

.agents/skills/b/b-add-documentation/SKILL.md

@@ -0,0 +1,22 @@
+---
+name: b-add-documentation
+description: Add comprehensive documentation for current code/feature per project standards (README, technical docs site, or inline comments). Use when the user types /b-add-documentation.
+disable-model-invocation: true
+---
+
+Add documentation for the current code/feature. Follow `.cursor/rules/base/docs.mdc` and `.cursor/rules/base/readme.mdc`.
+
+## Where to write
+
+| Layer | Role | Update when |
+| --- | --- | --- |
+| Technical docs (path in `AGENTS.md`) | Architecture, ADRs, how-to | Behavior, architecture, commands, conventions, or workflow changed |
+| `.cursor/rules` | Short constraints | A convention the agent must not violate changed |
+| Nearest README | How to run this app/package; links only | Scripts, setup, or package purpose changed |
+
+Inline comments only when the code is otherwise misleading. Do not copy MDX into rules or READMEs. Do not `@`-attach MDX.
+
+1. **Identify the topic**: Matching docs section, then existing topic page (create a page only if none fits)
+2. **Write or patch MDX** if the canonical explanation changed
+3. **Patch README** only for run/setup/scripts; link the MDX
+4. **Patch the glob-matched `.mdc`** only if a constraint changed

.agents/skills/workflow/add-error-handling/SKILL.md → .agents/skills/b/b-add-error-handling/SKILL.md

@@ -1,15 +1,22 @@
 ---
-name: add-error-handling
-description: Implement comprehensive error handling for current code to make it robust and resilient while maintaining good UX. Use when the user types /add-error-handling.
+name: b-add-error-handling
+description: Implement comprehensive error handling for current code to make it robust and resilient while maintaining good UX. Use when the user types /b-add-error-handling.
 disable-model-invocation: true
 ---
 
+## Purpose
+
 Implement comprehensive error handling for current code to make it robust and resilient while maintaining good UX.
 
+## Steps
+
 1. **Error Detection**: Identify potential failure points, edge cases, unhandled exceptions, missing validation, async/network call issues
 2. **Error Handling Strategy**: Implement try-catch blocks, add input validation/sanitization, create meaningful error messages/logging, design graceful degradation
 3. **Recovery Mechanisms**: Implement retry logic for transient failures, add fallback options, create circuit breakers, design proper error propagation
 4. **User Experience**: Provide clear error messages, implement proper error status codes for APIs, add loading states/error boundaries for UI, include helpful suggestions
 
-Follow `apps/docu/content/docs/architecture/error-handling.mdx`.
+Follow the error-handling page listed from the technical docs path in `AGENTS.md`.
+
+## Completion
 
+Read [completion evidence](../references/completion.md) before reporting completion.

.agents/skills/workflow/audit-accessibility/SKILL.md → .agents/skills/b/b-audit-accessibility/SKILL.md

@@ -1,14 +1,21 @@
 ---
-name: audit-accessibility
-description: Perform comprehensive accessibility audit of current UI code to ensure compliance with WCAG guidelines and provide inclusive user experience. Use when the user types /audit-accessibility.
+name: b-audit-accessibility
+description: Perform comprehensive accessibility audit of current UI code to ensure compliance with WCAG guidelines and provide inclusive user experience. Use when the user types /b-audit-accessibility.
 disable-model-invocation: true
 ---
 
+## Purpose
+
 Perform comprehensive accessibility audit of current UI code to ensure compliance with WCAG guidelines and provide inclusive user experience.
 
+## Steps
+
 1. **WCAG Compliance**: Check conformance to WCAG 2.1 guidelines (A, AA, AAA levels), verify proper semantic HTML structure, ensure keyboard navigation support, review color contrast/visual accessibility
 2. **Screen Reader Support**: Validate ARIA labels/descriptions, check heading hierarchy/structure, ensure form labels/error messages accessible, review dynamic content announcements
 3. **Interactive Elements**: Verify focus management/visible focus indicators, check tab order/keyboard shortcuts, ensure interactive elements properly sized, review modal/dialog accessibility
 4. **Testing & Tools**: Suggest automated accessibility testing tools, provide manual testing procedures, create accessibility test cases, recommend browser extensions/validators
 5. **Remediation**: Provide specific code fixes for each issue, include ARIA attributes/semantic improvements, suggest alternative approaches for complex interactions, create accessible component patterns
 
+## Completion
+
+Read [completion evidence](../references/completion.md) before reporting completion.

.agents/skills/b/b-build/SKILL.md

@@ -0,0 +1,30 @@
+---
+name: b-build
+description: Implement an agreed task incrementally and verify the result. Use when the user types /b-build.
+disable-model-invocation: true
+---
+
+## Purpose and inputs
+
+Use an existing plan or a sufficiently clear implementation request. Read the repository instructions, affected README/scripts, matching rules and skills, and technical docs. Resolve missing consequential decisions through the matching FIRST station.
+
+## Steps
+
+1. Inspect the working tree and identify the files owned by this task. State the acceptance conditions; preserve unrelated changes.
+2. Implement one complete slice using existing packages and patterns. Change owning schemas and run documented generators instead of editing generated clients or migrations.
+3. Run the smallest meaningful check for the changed behavior. A reproducible logic defect should have a regression check; use TDD when requested or required by the repository, not as a ritual for prose edits.
+4. Investigate failed checks before building dependent work. Separate regressions caused here from pre-existing or environmental failures; never weaken checks to obtain a pass.
+5. Review the diff and update matching technical docs and nearest README when behavior or conventions change. Update the product overlay only when product facts change.
+
+## Verification
+
+- [ ] The requested acceptance conditions are met.
+- [ ] Affected checks ran against the final change; any unverified behavior is named.
+- [ ] Sources, generated artifacts, and documentation agree.
+- [ ] The diff preserves unrelated work and contains no temporary instrumentation added here.
+
+## Handoff
+
+Return changed behavior, evidence, and remaining blockers. Commit, push, PR creation, and deployment require the user's request; build alone does not request them.
+
+Read [completion evidence](../references/completion.md) before reporting completion.

.agents/skills/b/b-clarify-task/SKILL.md

@@ -0,0 +1,23 @@
+---
+name: b-clarify-task
+description: Resolve consequential missing requirements from inspected context. Use when the user types /b-clarify-task.
+disable-model-invocation: true
+---
+
+## Purpose and inputs
+
+Use when missing requirements prevent a sound plan. Read the task, relevant docs and implementation before asking what the repository already answers.
+
+## Steps
+
+1. Separate known facts, reversible assumptions, and decisions that belong to the user.
+2. Ask only questions whose answers change scope, architecture, acceptance criteria, or a consequential boundary. Offer concise choices when useful; avoid a fixed interview quota.
+3. Incorporate answers into the existing task or plan. Continue independent work while awaiting an answer where the harness supports it.
+
+## Verification and handoff
+
+- [ ] The remaining questions affect a concrete decision.
+- [ ] Accepted answers and assumptions are reflected in the next step.
+- [ ] An already-authorized implementation is not delayed by a second generic confirmation.
+
+Continue the requested planning or implementation once its required inputs are available.

.agents/skills/b/b-code-review/SKILL.md

@@ -0,0 +1,29 @@
+---
+name: b-code-review
+description: Review a change for evidenced defects and risks without editing it. Use when the user types /b-code-review.
+disable-model-invocation: true
+---
+
+## Purpose and inputs
+
+Use the specified diff, branch, or PR and its intended behavior. Establish the review base and inspect existing tests and surrounding implementation. Remain read-only unless fixes are also requested.
+
+## Steps
+
+1. Understand the trigger and expected result before judging the implementation. Read tests to learn which claims are actually exercised.
+2. Check the dimensions below, concentrating on changed behavior and affected callers. Read domain rules when a boundary changes.
+3. Validate suspected defects with a concrete execution path, failing case, or source evidence. Do not invent timings or vulnerabilities from appearances.
+4. Order findings by impact. Give file/line, trigger, consequence, and smallest useful remedy. Separate blockers from optional suggestions.
+
+## Review checklist
+
+- [ ] Correctness: success, empty, invalid, failure, and concurrency cases relevant to this change.
+- [ ] Clarity: names and control flow expose intent; refactoring removes complexity instead of moving it.
+- [ ] Architecture: ownership, dependencies, generated sources, and existing shared helpers remain coherent.
+- [ ] Security: changed input and authorization boundaries, secret exposure, and denied access paths are checked.
+- [ ] Performance: changed queries, list bounds, repeated work, and client rendering have evidence appropriate to their risk.
+- [ ] Verification: tests exercise behavior and the reported checks cover the actual change.
+
+## Handoff
+
+Report actionable findings first, then unresolved questions and verification limits. If none are found, say so with the review scope and residual risks. An inspection is not proof that tests passed and is not approval to merge.

.agents/skills/workflow/coderabbit/SKILL.md → .agents/skills/b/b-coderabbit/SKILL.md

@@ -1,15 +1,22 @@
 ---
-name: coderabbit
-description: Fetch CodeRabbit review comments for current PR, analyze all issues, apply fixes automatically, and commit changes. Use when the user types /coderabbit.
+name: b-coderabbit
+description: Fetch CodeRabbit review comments for current PR, analyze all issues, apply fixes automatically, and commit changes. Use when the user types /b-coderabbit.
 disable-model-invocation: true
 ---
 
+## Purpose
+
 Fetch CodeRabbit review comments for current PR, analyze all issues, apply fixes automatically, and commit changes. Integrates CodeRabbit's AI code review directly into workflow.
 
+## Steps
+
 1. **Identify PR context**: Get current branch name, determine associated GitHub PR (if exists), if no PR exists check for uncommitted changes to review
 2. **Fetch CodeRabbit review**: Use CodeRabbit MCP to fetch review comments for PR, if no PR exists create review context from current changes, group comments by file/severity (critical/high/medium/low)
 3. **Analyze and prioritize**: Review each CodeRabbit comment for context/reasoning, categorize issues (bugs/security/performance/style/documentation), prioritize critical/high-severity issues first, note issues requiring clarification or cannot be auto-fixed
 4. **Apply fixes**: Fix issues file by file starting with highest priority, follow project coding standards/rules (see `.cursor/rules/`), ensure fixes address root cause not symptoms, run linting after each significant change: `pnpm lint:fix`, verify fixes don't introduce new issues
 5. **Verify changes**: Run linting: `pnpm lint` (skip if only markdown files changed), check for compilation errors, ensure tests still pass (if applicable), review diff to confirm all issues addressed
 6. **Commit fixes**: Stage all fixed files, create commit with descriptive message: `fix: address CodeRabbit review comments`, include summary of fixes applied, reference specific issues if helpful
 
+## Completion
+
+Read [completion evidence](../references/completion.md) before reporting completion.

.agents/skills/workflow/council/SKILL.md → .agents/skills/b/b-council/SKILL.md

@@ -1,12 +1,19 @@
 ---
-name: council
-description: Based on the given area of interest, dig around the codebase to gather information, spawn multiple task agents for deeper exploration with variance, then use the collected information to accomplish. Use when the user types /council.
+name: b-council
+description: Based on the given area of interest, dig around the codebase to gather information, spawn multiple task agents for deeper exploration with variance, then use the collected information to accomplish. Use when the user types /b-council.
 disable-model-invocation: true
 ---
 
+## Purpose
+
 Based on the given area of interest, dig around the codebase to gather information, spawn multiple task agents for deeper exploration with variance, then use the collected information to accomplish what the user wants.
 
+## Steps
+
 1. **Gather information**: Dig around the codebase in terms of given area of interest, gather general information such as keywords and architecture overview
 2. **Spawn task agents**: Spawn off n=10 (unless specified otherwise) task agents to dig deeper into the codebase, some should be out of the box for variance
 3. **Use information**: Once the task agents are done, use the information to do what the user wants (if user is in plan mode, create the plan per @.cursor/rules/base/general.mdc: References, assumptions, deferrals)
-4. **Persist**: If the user asked for a plan, keep it in the Cursor plan. If the finding changes product facts, patch the Product instance in `_first/FIRST.md` (Basilic: `_first/basilic/PRODUCT.md`). If it changes quality or workflow facts, patch the matching MDX or FIRST overlay. Not `__dev/` as Fact. Issues for later work go to GitHub Issues.
+
+## Completion
+
+Read [completion evidence](../references/completion.md) before reporting completion.

.agents/skills/b/b-debug-browser/SKILL.md

@@ -0,0 +1,24 @@
+---
+name: b-debug-browser
+description: Reproduce and resolve an authorized browser issue using runtime evidence. Use when the user types /b-debug-browser.
+disable-model-invocation: true
+---
+
+## Purpose and inputs
+
+Use the affected URL, expected interaction, and available browser tools. Inspect the app's README and browser-testing conventions first.
+
+## Steps
+
+1. Reproduce the interaction and collect relevant DOM, network, and console evidence before adding instrumentation.
+2. Trace the failure to its owning client or server boundary. Use the repository logger for necessary temporary traces; avoid secrets in captured output.
+3. Apply the smallest authorized fix, then replay the same interaction and relevant failure or empty states.
+4. Run affected automated checks and remove temporary traces introduced here. If tools or credentials block reproduction, report the exact limitation rather than looping without new evidence.
+
+## Verification and handoff
+
+- [ ] Original interaction succeeds in the browser after the fix.
+- [ ] Relevant failures, loading states, and console/network errors were inspected.
+- [ ] Reported runtime evidence and automated results are distinguished.
+
+Return the cause, changed behavior, and evidence. Do not ask the user to test something the available tools can verify.