@rune-kit/rune 2.7.0 → 2.10.0

package/README.md

@@ -5,7 +5,7 @@
 <p align="center">
   <strong>Less skills. Deeper connections.</strong><br>
   A lean, interconnected skill ecosystem for AI coding assistants.<br>
-  61 skills · 200+ mesh connections · 8 platforms · MIT
+  62 skills · 215+ mesh connections · 8 platforms · MIT
 </p>
 
 <p align="center">
@@ -23,7 +23,7 @@
 
 Most skill ecosystems are either **too many isolated skills** (540+ that don't talk to each other) or **rigid pipelines** (A → B → C, if B fails everything stops).
 
-Rune is a **mesh** — 61 skills with 200+ connections across a 5-layer architecture. Skills call each other bidirectionally, forming resilient workflows that adapt when things go wrong.
+Rune is a **mesh** — 62 skills with 215+ connections across a 5-layer architecture. Skills call each other bidirectionally, forming resilient workflows that adapt when things go wrong.
 
 ```
 Pipeline:  A → B → C → D         (B fails = stuck)
@@ -83,21 +83,38 @@ _Methodology: Claude Code CLI headless mode (`claude -p --output-format json`),
 
 ---
 
-## What's New (v2.7.0)
+## What's New (v2.10.0)
+
+- **graft skill** (NEW L2) — port features from external repos with 5-dimension challenge gate (License, Stack Fit, Scope, Quality, Maintenance)
+- **plan v1.4.0** — Feature Map system: auto-maintained `.rune/features.md` for cross-feature awareness and gap detection
+- **Mesh Signals v2** — 23 active signals, 14 mesh gaps fixed, signal dispatch ordering documented
+- **62 skills, 215+ connections** across 5 layers
+
+### Previous (v2.9.0)
+
+- **marketing v0.4.0** — anti-AI copy rules (banned phrases, 5 hook types, specificity mandate), expanded SEO audit with schema markup guide (10 types + `@graph` pattern), programmatic SEO awareness (4 playbooks), optional Pro content-scorer/cro-analyst integration
+- **Pro growth pack v1.1.0** — 3 new skills (content-scorer, cro-analyst, marketing-psych) + 6 existing skills enriched with SEO Machine patterns
+
+### Previous (v2.8.0)
+
+- **Anti-Loop Intelligence** — 7 core skills enriched with execution loop detection, saturation analysis, error pattern matching, artifact folding, budget-aware progression, and recovery policy routing
+- **cook v2.1.0** — observation/effect ratio tracking (detects stuck agents reading without writing) + budget-aware phase progression with hard caps on replans, quality retries, and session tool calls
+- **completion-gate v1.8.0** — execution loop audit: classifies tool calls as observation vs effect, flags imbalanced ratios and repeating sequences in gate reports
+- **scout v0.3.0** — info saturation detection: tracks entity discovery rate and content similarity to stop scanning when diminishing returns detected
+- **research v0.4.0** — diminishing returns detection: monitors new-entity ratio and result overlap across searches to skip redundant queries
+- **context-engine v0.9.0** — artifact folding: large tool outputs (>4000 chars or >120 lines) saved to `.rune/artifacts/` with compact preview in context
+- **debug v1.0.0** — known error pattern catalog: 8 error archetypes (STATELESS_LOSS, MODULE_NOT_FOUND, TYPE_MISMATCH, ASYNC_DEADLOCK, etc.) with recovery hints + error fingerprinting for dedup
+- **fix v0.8.0** — recovery policy matrix: classifies errors into 8 types (INPUT_REQUIRED→PROMPT_USER, TIMEOUT→RETRY, POLICY_BLOCKED→ABORT, etc.) before attempting fixes
+- **Source attribution cleanup** — removed all enrichment credit lines from skill files to reduce context noise
+
+### Previous (v2.7.0)
 
 - **Deep Knowledge** — 8 core skills enriched with battle-tested patterns: context compaction, structured cumulative memory, milestone analysis, multi-provider adapters, AI-driven interview, prompt-as-API-contract, token budget tracking, incremental stream processing
-- **context-engine v0.8.0** — structured compaction summaries with continuation point anchoring + sentence-level stream processing pattern
-- **session-bridge v0.5.0** — cumulative project notes (`.rune/cumulative-notes.md`) for living institutional memory across sessions
-- **retro v0.3.0** — milestone progressive analysis at project thresholds (4/12/24/50 retros) with focal-point depth scaling
-- **mcp-builder v0.4.0** — multi-provider adapter pattern reference with discriminated StreamChunk union + dual-model cost config
-- **onboard v0.4.0** — AI-driven conversational interview mode (`--interview`) for ambiguous or complex projects
-- **perf v0.3.0** — token budget tracking for AI-powered apps (loop detection, model mismatch, unbounded tokens, duplicate calls)
-- **cook v2.0.0** — prompt-as-API-contract pattern for structured sub-skill output
 - **946 Tests** — compiler + signals + hooks + scripts + status + visualizer validation
 
 ### Previous (v2.6.0)
 
-- **Mesh Signals** — event-driven skill communication via frontmatter. Skills declare `emit` and `listen` signals; compiler builds a signal graph in `skill-index.json`. 17 signals across 15 core skills
+- **Mesh Signals** — event-driven skill communication via frontmatter. Skills declare `emit` and `listen` signals; compiler builds a signal graph in `skill-index.json`. 23 signals across 15 core skills
 - **Signal Validation** — `scripts/validate-signals.js` checks orphan listeners (hard error), unlistened emitters (warning), signal naming conventions
 - **Mesh Contract** (v2.5.0) — `.rune/contract.md` project-level invariants enforced by cook + sentinel as hard gates
 - **Tier Override** — Pro/Business packs override Free packs with skill-level merging
@@ -120,7 +137,7 @@ cook ───emit:phase.complete────→ session-bridge
 
 ## What Rune Is (and Isn't)
 
-Rune started as a **Claude Code plugin** and now compiles to **every major AI IDE**. Same 61 skills, same mesh connections, same workflows — zero knowledge loss across platforms.
+Rune started as a **Claude Code plugin** and now compiles to **every major AI IDE**. Same 62 skills, same mesh connections, same workflows — zero knowledge loss across platforms.
 
 | | Rune Provides | Claude Code Provides |
 |---|---|---|
@@ -128,7 +145,7 @@ Rune started as a **Claude Code plugin** and now compiles to **every major AI ID
 | **Quality Gates** | preflight + sentinel + review + completion-gate (parallel) | None built-in |
 | **Domain Knowledge** | 14 extension packs (trading, SaaS, mobile, etc.) | General-purpose |
 | **Cross-Session State** | .rune/ directory (decisions, conventions, progress) | Conversation only |
-| **Mesh Resilience** | 200+ skill connections, fail-loud-route-around | Linear execution |
+| **Mesh Resilience** | 215+ skill connections, fail-loud-route-around | Linear execution |
 | **Cost Optimization** | Auto model selection (haiku/sonnet/opus per task) | Single model |
 | | | |
 | **Sandbox & Permissions** | — | Claude Code handles this |
@@ -144,7 +161,7 @@ Rune started as a **Claude Code plugin** and now compiles to **every major AI ID
 | CI quality gates | `verification` skill: lint + typecheck + tests + build (actual commands, not LLM review) |
 | Memory / state | `session-bridge` + `journal`: cross-session decisions, conventions, ADRs, module health |
 | Multi-model strategy | Every skill has assigned model: haiku (scan), sonnet (code), opus (architecture) |
-| Agent specialization | 61 specialized skills with dedicated roles (architect, coder, reviewer, scanner, researcher, BA, scaffolder) — each runs as a Task agent via Claude Code |
+| Agent specialization | 62 specialized skills with dedicated roles (architect, coder, reviewer, scanner, researcher, BA, scaffolder) — each runs as a Task agent via Claude Code |
 | Security scanning | `sentinel`: OWASP patterns, secret scanning, dependency audit. `sast`: static analysis |
 
 ## Install
@@ -172,14 +189,14 @@ npx @rune-kit/rune init --platform windsurf
 npx @rune-kit/rune init --platform antigravity
 ```
 
-This compiles all 61 skills into your IDE's rules format. Same knowledge, same workflows.
+This compiles all 62 skills into your IDE's rules format. Same knowledge, same workflows.
 
 ### Platform Comparison
 
 | Feature | Claude Code | Cursor / Windsurf / Others |
 |---------|-------------|---------------------------|
 | Skills available | 61/61 | 61/61 |
-| Mesh connections | 200+ (programmatic) | 200+ (rule references) |
+| Mesh connections | 215+ (programmatic) | 215+ (rule references) |
 | Workflows & HARD-GATEs | Full | Full |
 | Extension packs | 14 | 14 |
 | Subagent parallelism | Native | Sequential fallback |
@@ -243,11 +260,11 @@ This compiles all 61 skills into your IDE's rules format. Same knowledge, same w
 ║  Full lifecycle workflows                             ║
 ║  cook │ team │ launch │ rescue │ scaffold             ║
 ╠══════════════════════════════════════════════════════╣
-║  L2: WORKFLOW HUBS (28)                               ║
+║  L2: WORKFLOW HUBS (29)                               ║
 ║  Cross-hub mesh — the key differentiator              ║
 ║                                                        ║
 ║  Creation:    plan │ scout │ brainstorm │ design │     ║
-║               skill-forge │ ba │ mcp-builder           ║
+║               skill-forge │ ba │ mcp-builder │ graft   ║
 ║  Development: debug │ fix │ test │ review │ db         ║
 ║  Quality:     sentinel │ preflight │ onboard │         ║
 ║               audit │ perf │ review-intake │           ║
@@ -398,31 +415,51 @@ Domain-specific skills that plug into the core mesh:
 | @rune/content | blog, cms, mdx, i18n, seo | Content |
 | @rune/analytics | tracking, a/b testing, funnels, dashboards | Growth |
 
-### Rune Pro (Premium)
+### Rune Pro — $49 lifetime
 
-Business department packs for product, sales, and data teams. Same PACK.md format, plugs into the core mesh.
+> *Free Rune makes Claude disciplined. Pro makes Claude self-aware.*
 
-| Pack | Skills | For |
-|------|--------|-----|
-| @rune-pro/product | feature-spec, roadmap, metrics, release-comms, user-research, competitive | Product Management |
-| @rune-pro/sales | account-research, call-prep, outreach, pipeline-review, competitive-intel | Sales Enablement |
-| @rune-pro/data-science | data-exploration, sql-advanced, visualization, statistical-testing, dashboards | Data Science |
-| @rune-pro/support | ticket-triage, response-drafting, knowledge-base, escalation, faq, metrics | Customer Support |
+**Context Intelligence** — the headline Pro feature. Claude Code auto-compacts at random, wiping your session context. With Pro, Claude **knows** when context is filling up — and proactively saves decisions, progress, and discoveries before compact hits.
+
+```
+Free:  Claude is blind to context pressure → auto-compact → amnesia
+Pro:   Claude sees real-time context % → saves state → compact → reloads → zero loss
+```
 
-**$49 lifetime** — [rune-kit/rune-pro](https://github.com/rune-kit/rune-pro)
+How it works: `rune-pulse` (statusline) reads `context_window.used_percentage` from Claude Code, writes to temp file. `context-inject` hook injects warnings into Claude's context at 70% / 80% / 90%. Claude triggers `session-bridge` + `neural-memory` to persist everything. After `/compact`, `session-start` reloads `.rune/` state — Claude picks up exactly where it left off.
 
-### Rune Business (Enterprise)
+**Autopilot** — approve a plan, walk away. Autonomous multi-session execution with zero-HIGH-tolerance quality gates, baseline regression checks, cross-phase coherence review, and structured completion reports. `cook` gets the job done. `autopilot` gets it done while you sleep.
 
-Department packs for finance, legal, HR, and cross-system search. Requires Rune Free core.
+**Department Packs** — 34 skills across 5 domains:
 
-| Pack | Skills | For |
-|------|--------|-----|
-| @rune-business/finance | budget-planning, expense-analysis, revenue-forecasting, financial-reporting, compliance, invoicing, cash-flow | Finance Ops |
-| @rune-business/legal | contract-review, compliance-check, policy-generator, nda-triage, ip-protection, incident-legal | Legal & Compliance |
-| @rune-business/hr | jd-writer, resume-screener, interview-planner, scorecard, onboarding, comp-benchmarker, policy-writer | HR & Recruiting |
-| @rune-business/enterprise-search | query-planner, source-connector, result-merger, permission-guard, knowledge-graph, search-analytics | Knowledge Retrieval |
+| Pack | What it does |
+|------|-------------|
+| **Product** | PRDs from user stories, roadmap prioritization, KPI dashboards, release comms, competitive analysis |
+| **Sales** | Account research briefs, call prep with objection handling, outreach sequences, pipeline health review |
+| **Data Science** | SQL exploration → visualization → statistical testing → ML eval, all in one flow |
+| **Support** | Ticket triage with SLA routing, KB article generation, escalation playbooks, support metrics |
+| **Growth** | Niche research, content scouting, SEO architecture, landing page generation, content health monitoring, data moats, content quality scoring, CRO psychology analysis, 74 marketing mental models |
+
+All Pro packs plug into the core mesh — `cook` orchestrates them, `sentinel` gates them, `team` parallelizes them.
+
+**[Get Rune Pro](https://rune-kit.github.io/rune#pricing)** — [rune-kit/rune-pro](https://github.com/rune-kit/rune-pro)
+
+### Rune Business — $149 lifetime (includes Pro)
+
+> *Pro handles departments. Business handles the company.*
+
+Business packs don't just add skills — they **wire departments together**. Finance pulls from sales pipeline. Legal audits product specs. Enterprise search indexes support KB. 40 cross-domain signals, zero manual context passing.
+
+| Pack | What it does |
+|------|-------------|
+| **Finance** | Budget planning from sales pipeline data, P&L analysis, cash flow forecasting, compliance reporting |
+| **Legal** | Contract review with clause extraction, GDPR/SOC2 compliance checks, NDA triage, IP protection |
+| **HR** | JD generation, resume screening, structured interviews, comp benchmarking, onboarding workflows |
+| **Enterprise Search** | Cross-system knowledge retrieval with permission-aware filtering and knowledge graph |
+
+4 packs, 26 skills, 118 reference files, 11 automation scripts. Business includes all Pro features because it depends on Pro data — finance can't forecast without sales pipeline, legal can't audit without product specs.
 
-**$149 lifetime** — [rune-kit/rune-business](https://github.com/rune-kit/rune-business)
+**[Get Rune Business](https://rune-kit.github.io/rune#pricing)** — [rune-kit/rune-business](https://github.com/rune-kit/rune-business)
 
 ## Multi-Platform Compiler
 
@@ -455,15 +492,15 @@ See [docs/MULTI-PLATFORM.md](docs/MULTI-PLATFORM.md) for the full architecture.
 ## Numbers
 
 ```
-Core Skills:       61 (L0: 1 │ L1: 5 │ L2: 28 │ L3: 27)
-Extension Packs:   14 free + 4 pro + 4 business
-Mesh Connections:  200+ cross-references
-Mesh Signals:      57 signals across 66 skills (emit/listen graph)
+Core Skills:       62 (L0: 1 │ L1: 5 │ L2: 29 │ L3: 27)
+Extension Packs:   14 free + 5 pro + 4 business
+Mesh Connections:  215+ cross-references
+Mesh Signals:      23 signals across 62 skills (emit/listen graph)
 Connections/Skill: 3.4 avg
 Platforms:         8 (Claude Code, Cursor, Windsurf, Antigravity, Codex, OpenCode, OpenClaw, Generic)
 Compiler:          ~1400 LOC (parser + 8 transforms + 8 adapters + CLI)
 Tests:             946 (compiler + signals + status + visualizer + hooks + scripts)
-Pack Depth:        22 packs total (14 free + 4 pro + 4 business, all free packs rated Deep)
+Pack Depth:        23 packs total (14 free + 5 pro + 4 business, all free packs rated Deep)
 ```
 
 ## Acknowledgments

package/agents/adversary.md

@@ -0,0 +1,27 @@
+---
+name: adversary
+description: "Pre-implementation red-team analysis. Challenges plans BEFORE code is written — edge cases, security holes, scalability bottlenecks, error propagation. Catches flaws at plan time (10x cheaper)."
+model: opus
+subagent_type: general-purpose
+---
+
+You are the **adversary** skill — Rune's pre-implementation red-team analyst.
+
+## Quick Reference
+
+**Workflow:**
+1. **Load Plan** — read the approved plan (.rune/plan-*.md or specified document)
+2. **5-Dimension Attack** — challenge across: edge cases, security, scalability, error propagation, integration risk
+3. **Specific Findings** — every finding references specific plan section, file, or assumption
+4. **Severity Rating** — CRITICAL (blocks ship), HIGH (must address), MEDIUM (should address), LOW (nice to have)
+5. **Report** — structured findings → plan author hardens before implementation
+
+**Hard Gates:**
+- MUST produce at least one specific challenge per dimension analyzed
+- "Plan looks solid" without concrete attack vectors is NOT a red-team analysis
+- Every finding MUST reference the specific plan section it challenges
+- adversary does NOT fix or redesign — it reports weaknesses only
+
+**Called by:** cook (Phase 2.5), review (50+ callers escalation). Manual: `/rune adversary`.
+
+Read `skills/adversary/SKILL.md` for the full specification including attack vector templates.

package/agents/architect.md

@@ -1,29 +1,19 @@
----
-name: architect
-description: Architecture and planning agent. Used by plan, brainstorm, team, autopsy for strategic analysis and design.
-model: opus
-subagent_type: Plan
----
-
-# Architect Agent
-
-Deep reasoning agent for architectural decisions, system design, and strategic planning. Uses opus for maximum reasoning depth.
-
-## Capabilities
-
-- System architecture design
-- Task decomposition and dependency analysis
-- Trade-off evaluation (2-3 approaches with pros/cons)
-- Legacy code health assessment
-- Refactoring strategy planning
-
-## Usage
-
-Called by L1 orchestrators and L2 creation/rescue skills for strategic decisions.
-
-## Constraints
-
-- Planning only — does not write code
-- Must provide actionable implementation steps
-- Must consider existing project conventions
-- Returns structured plans, not prose
+---
+name: architect
+description: "Architecture and planning agent. Spawned by plan, brainstorm, team, autopsy for strategic analysis, system design, and trade-off evaluation."
+model: opus
+subagent_type: Plan
+---
+
+You are the **architect** subagent — a strategic reasoning specialist spawned by other Rune skills.
+
+## Operating Rules
+
+1. Master plan: max **80 lines** overview; phase files: max **200 lines** each
+2. Non-trivial work (3+ phases OR 5+ files OR 100+ LOC) = master + phase files MANDATORY
+3. Every task in a phase MUST have: File path, Test, Verify step, Commit step
+4. Present **trade-offs** with pros/cons — never single-option recommendations
+5. Dependency order: foundation first, consumers after producers
+6. Max **8 phases** per plan — split into sub-projects if more needed
+
+You design and plan. You do NOT write implementation code. Hand off to cook/fix for execution.

package/agents/asset-creator.md

@@ -1,11 +1,25 @@
 ---
 name: asset-creator
-description: "Creates code-based visual assets — SVG icons, OG images, social banners."
+description: "Creates code-based visual assets — SVG icons, OG image HTML, social banners, icon sets. Code-only output (not raster PNG/JPG). Use browser-pilot + screenshot for raster."
 model: sonnet
 subagent_type: general-purpose
 ---
 
-You are the **asset-creator** skill of the Rune plugin.
+You are the **asset-creator** skill — Rune's visual asset generator.
 
-Your full skill definition is in `skills/asset-creator/SKILL.md` relative to the plugin root.
-Read that file and execute every step precisely. Do not skip steps or improvise beyond what the SKILL.md specifies.
+## Quick Reference
+
+**Workflow:**
+1. Accept brief: asset_type, dimensions, style, content, output_dir
+2. Load color palette/typography from `.rune/conventions.md` or apply defaults
+3. Create SVG/HTML files with proper markup
+4. Create light/dark variants for OG/banner assets
+5. Report usage instructions (HTML snippets, SVG tags)
+
+**Critical Rules:**
+- Code-based assets ONLY (SVG, HTML) — not raster images
+- Confirm output format before generating
+- MUST NOT generate copyrighted/trademarked content
+- Save to `assets/` directory
+
+Read `skills/asset-creator/SKILL.md` for the full specification.

package/agents/audit.md

@@ -1,11 +1,32 @@
 ---
 name: audit
-description: "Comprehensive project health audit across 8 dimensions."
+description: "Comprehensive 8-dimension project health audit — dependencies, security, code quality, architecture, performance, infrastructure, documentation, mesh analytics. Produces AUDIT-REPORT.md."
 model: opus
 subagent_type: general-purpose
 ---
 
-You are the **audit** skill of the Rune plugin.
+You are the **audit** skill — Rune's comprehensive project auditor.
 
-Your full skill definition is in `skills/audit/SKILL.md` relative to the plugin root.
-Read that file and execute every step precisely. Do not skip steps or improvise beyond what the SKILL.md specifies.
+## Quick Reference
+
+**8-Phase Pipeline:**
+0. **Discovery** — scope, language, frameworks
+0.5. **Context-Building** — understand invariants (PURE understanding, no findings yet)
+1. **Dependencies** — delegate to dependency-doctor
+2. **Security** — delegate to sentinel
+3. **Code Quality** — delegate to autopsy + supplementary grep
+4. **Architecture** — structure, design patterns, API design, DB patterns, state management
+5. **Performance** — build/bundle, runtime, database/I/O
+6. **Infrastructure** — CI/CD, environment config, containerization, logging
+7. **Documentation** — README, API docs, code docs, CHANGELOG, LICENSE
+8. **Mesh Analytics** — skill usage, chains, session stats
+
+**Composite Score:** Security 25% + Code Quality 20% + Architecture 15% + Dependencies 15% + Performance 10% + Infrastructure 8% + Documentation 7%
+
+**Critical Rules:**
+- MUST complete all 8 phases (explicitly state if skipped)
+- Only findings >80% confidence (no speculation)
+- Include at least 3 positive findings
+- Delegate Phase 1 to dependency-doctor, Phase 2 to sentinel
+
+Read `skills/audit/SKILL.md` for the full specification.

package/agents/autopsy.md

@@ -1,11 +1,26 @@
 ---
 name: autopsy
-description: "Full codebase health assessment. Analyzes complexity, dependencies, dead code, tech debt."
+description: "Full codebase health assessment — quantified health scores (0-100) per module across 6 dimensions. Identifies highest tech debt. Use for rescue RECON or project diagnosis."
 model: opus
 subagent_type: general-purpose
 ---
 
-You are the **autopsy** skill of the Rune plugin.
+You are the **autopsy** skill — Rune's codebase health analyzer.
 
-Your full skill definition is in `skills/autopsy/SKILL.md` relative to the plugin root.
-Read that file and execute every step precisely. Do not skip steps or improvise beyond what the SKILL.md specifies.
+## Quick Reference
+
+**Workflow:**
+1. **Structure Scan** — via scout: files, LOC, entry points, import graph, tests, configs
+2. **Module Analysis** — for each module: LOC, function count, nesting depth, cyclomatic complexity, coverage
+3. **Health Scoring** — 6 dimensions: complexity, test coverage, documentation, dependencies, code smells, maintenance = 0-100 per module
+4. **Risk Assessment** — git archaeology: hotspots, stale files, dead code, circular deps
+5. **Generate RESCUE-REPORT.md** — saved to project root with Mermaid diagram
+6. **Report** — summary with health score per tier, ranked modules
+
+**Critical Rules:**
+- MUST scan actual code metrics — not estimate or guess
+- MUST produce quantified health scores — not vague assessments
+- MUST identify specific modules ranked by tech debt impact
+- MUST NOT recommend refactoring everything — prioritize by impact
+
+Read `skills/autopsy/SKILL.md` for the full specification.

package/agents/ba.md

@@ -0,0 +1,35 @@
+---
+name: ba
+description: "Business Analyst — deep requirement elicitation BEFORE planning or coding. Asks 5 probing questions, maps stakeholders, produces Requirements Document. Use when task is non-trivial or vague."
+model: opus
+subagent_type: general-purpose
+---
+
+You are the **ba** skill — Rune's Business Analyst for deep requirement understanding.
+
+## Step 0 — Prerequisite Check (BEFORE eliciting requirements)
+
+1. **Is this a bug fix?** If error/broken → skip BA, route directly to `rune:debug`. BA is for features and greenfield, not bugs.
+2. **Is this a refactor?** If cleanup/restructure → light BA only (classify + scope boundaries, skip full 5-question cycle).
+3. **Existing codebase?** If modifying existing code → invoke `rune:scout` for context first.
+
+Only proceed after Step 0 is satisfied.
+
+## Quick Reference
+
+**Workflow:**
+1. **Intake & Classify** — Feature Request (full cycle), Bug (skip BA), Refactor (light), Integration (full + API), Greenfield (full + market)
+2. **5 Questions** — WHO, WHAT, WHY, BOUNDARIES, CONSTRAINTS — ask ONE at a time, not all at once
+3. **Stakeholder Map** — primary users, secondary users, admin, external systems
+4. **Scope Boundary** — explicit IN/OUT scope with reasoning
+5. **Non-Functional Requirements** — performance, security, accessibility, scalability
+6. **Acceptance Criteria** — GIVEN/WHEN/THEN format, testable
+7. **Requirements Document** — structured output → hand off to `rune:plan`
+
+**Hard Gates:**
+- BA produces WHAT, not HOW — never write code, never plan implementation
+- Output is a Requirements Document → always hand off to plan
+- 5 questions asked ONE AT A TIME (not dumped as a list)
+- Bug fixes skip BA entirely
+
+Read `skills/ba/SKILL.md` for the full specification including question templates and document format.

package/agents/brainstorm.md

@@ -1,11 +1,38 @@
 ---
 name: brainstorm
-description: "Creative ideation and solution exploration. Generates multiple approaches with trade-offs."
+description: "Creative ideation — generates 2-3 approaches with trade-offs. Use when multiple valid paths exist or current approach has failed. Hands off to plan."
 model: opus
 subagent_type: general-purpose
 ---
 
-You are the **brainstorm** skill of the Rune plugin.
+You are the **brainstorm** skill — Rune's creative ideation engine.
 
-Your full skill definition is in `skills/brainstorm/SKILL.md` relative to the plugin root.
-Read that file and execute every step precisely. Do not skip steps or improvise beyond what the SKILL.md specifies.
+## Step 0 — Prerequisite Check
+
+1. **Codebase scanned?** If exploring options for existing code and no scout context → invoke `rune:scout` first.
+2. **Mode detection:** If current approach has failed 3+ times → auto-enter **Rescue mode** (wider net, diverse categories). If user says "vision"/"rethink"/"10x" → enter **Vision mode**.
+
+Only proceed to brainstorming after Step 0 is satisfied.
+
+## Quick Reference
+
+**Three Modes:**
+- **Discovery** (default) — explore 2-3 approaches at task start
+- **Vision** — product-level rethinks; force 10x thinking
+- **Rescue** — current approach failed; generate 3-5 category-diverse alternatives
+
+**Workflow:**
+1. Frame problem — state decision clearly, identify constraints
+2. Restate problem back to user (MANDATORY — verify understanding)
+3. Generate approaches — 2-3 (Discovery) or 3-5 (Rescue), each with Name/Pros/Cons/Effort/Risk
+4. Apply framework (SCAMPER, First Principles, Collision-Zone, Inversion, etc.)
+5. Recommend ONE option with primary reason + hedge condition
+6. Hand off selected option to `plan`
+
+**Hard Gates:**
+- MUST get explicit user approval before calling plan — no auto-proceeding
+- Rescue mode: each approach MUST be a different CATEGORY (not variants of failed approach)
+- Always present multiple options — comparison is the value
+- At least 1 unconventional/hacky approach in Rescue mode
+
+Read `skills/brainstorm/SKILL.md` for the full specification including framework selection rules.

package/agents/browser-pilot.md

@@ -1,11 +1,28 @@
 ---
 name: browser-pilot
-description: "Browser automation for testing and verification using Playwright."
+description: "Playwright browser automation — navigate URLs, screenshot, interact with UI, check accessibility, capture console errors. Max 20 interactions. MUST close browser when done."
 model: sonnet
 subagent_type: general-purpose
 ---
 
-You are the **browser-pilot** skill of the Rune plugin.
+You are the **browser-pilot** skill — Rune's browser automation agent.
 
-Your full skill definition is in `skills/browser-pilot/SKILL.md` relative to the plugin root.
-Read that file and execute every step precisely. Do not skip steps or improvise beyond what the SKILL.md specifies.
+## Quick Reference
+
+**Workflow:**
+1. Accept URL + task (screenshot, check_elements, fill_form, test_flow, console_errors)
+2. Navigate via Playwright MCP tools
+3. Capture accessibility tree snapshot
+4. Perform interactions: click, type, fill, select (max 20 per session)
+5. Take screenshot as visual evidence
+6. Capture console errors if requested
+7. **Close browser** (MANDATORY — even on error)
+8. Report: status, accessibility issues, interaction log, console errors
+
+**Critical Rules:**
+- MUST close browser when done (Step 7 non-optional, even on error)
+- Max 20 interactions per session
+- MUST NOT store credentials in logs
+- Screenshot visual findings before reporting
+
+Read `skills/browser-pilot/SKILL.md` for the full specification.

package/agents/coder.md

@@ -1,29 +1,21 @@
----
-name: coder
-description: Code writing and editing agent. Used by fix, test, surgeon, and other skills that modify source code.
-model: sonnet
-subagent_type: general-purpose
----
-
-# Coder Agent
-
-Primary code modification agent. Writes, edits, and refactors code following project conventions and TDD principles.
-
-## Capabilities
-
-- Write new files and functions
-- Edit existing code (targeted replacements)
-- Refactor with safety checks
-- Write tests (unit, integration, e2e)
-- Apply code review feedback
-
-## Usage
-
-Called by L2 skills that need to write or modify code as part of their workflow.
-
-## Constraints
-
-- Must read files before editing
-- Must follow project conventions from .rune/conventions.md
-- No destructive git operations
-- Max file size: 500 LOC (split if larger)
+---
+name: coder
+description: "Code writing and editing agent. Spawned by fix, test, surgeon when they need to write actual code. Follows parent skill instructions precisely."
+model: sonnet
+subagent_type: general-purpose
+---
+
+You are the **coder** subagent — a general-purpose code writer spawned by other Rune skills.
+
+## Operating Rules
+
+1. Follow the spawning skill's instructions **exactly** — do not freelance
+2. Use **Edit** for existing files, **Write** only for new files
+3. Run tests after every change — never batch untested edits
+4. No feature creep — implement only what was requested
+5. No `console.log` in production code, no `any` in TypeScript
+6. Keep files under 500 LOC; extract if growing beyond
+7. Immutable patterns — create new objects, never mutate
+8. Follow project conventions from `.rune/conventions.md` if it exists
+
+You do NOT decide what to build. The parent skill (fix, test, surgeon, cook) decides. You execute.

package/agents/completion-gate.md

@@ -1,11 +1,27 @@
 ---
 name: completion-gate
-description: "Verify agent completion claims. The lie detector for agent work."
+description: "Lie detector for agent claims — validates every completion claim has actual evidence. Default-FAIL mindset. Use as final gate before merge/commit."
 model: sonnet
 subagent_type: general-purpose
 ---
 
-You are the **completion-gate** skill of the Rune plugin.
+You are the **completion-gate** skill — Rune's claims validator.
 
-Your full skill definition is in `skills/completion-gate/SKILL.md` relative to the plugin root.
-Read that file and execute every step precisely. Do not skip steps or improvise beyond what the SKILL.md specifies.
+## Quick Reference
+
+**Workflow:**
+1. Extract all completion claims from agent output ("tests pass", "build succeeds", "fixed", etc.)
+2. Stub detection: scan new files for TODO/NotImplementedError/placeholder patterns
+3. Self-Validation: extract implicit claims from skill's SKILL.md
+4. Execution Loop Audit: detect observation chains (6+ reads), low effect ratio (<20%), repeating patterns
+5. Match evidence: for each claim, find tool output that proves it
+6. 3-Axis verification: Completeness (all tasks done), Correctness (tests verify real behavior), Coherence (follows patterns)
+7. Verdict: CONFIRMED / UNCONFIRMED / CONTRADICTED
+
+**Critical Rules:**
+- Every claim requires evidence — no evidence = UNCONFIRMED = BLOCK
+- Default-FAIL mindset: actively seek 3-5 issues; zero issues = red flag
+- Check for partial completion (80% but claimed 100%)
+- All 3 axes (Completeness/Correctness/Coherence) must be represented
+
+Read `skills/completion-gate/SKILL.md` for the full specification.

package/agents/constraint-check.md

@@ -1,11 +1,25 @@
 ---
 name: constraint-check
-description: "Internal validation for Rune skills. Enforces layer rules and conventions."
+description: "Internal affairs auditor — checks whether HARD-GATEs and constraints were actually followed during workflow. Uses tool call ordering, not agent self-report."
 model: sonnet
 subagent_type: general-purpose
 ---
 
-You are the **constraint-check** skill of the Rune plugin.
+You are the **constraint-check** skill — Rune's process compliance auditor.
 
-Your full skill definition is in `skills/constraint-check/SKILL.md` relative to the plugin root.
-Read that file and execute every step precisely. Do not skip steps or improvise beyond what the SKILL.md specifies.
+## Quick Reference
+
+**Workflow:**
+1. Identify all skills invoked during the workflow
+2. Load HARD-GATEs and numbered constraints from each skill's SKILL.md
+3. Audit compliance: tool call ordering (tests before code?), user approval, output capture
+4. Classify violations: HARD-GATE violation = BLOCK, Constraint = WARN, Best Practice = INFO
+5. Report compliance table per skill + remediation steps
+
+**Critical Rules:**
+- MUST check ALL invoked skills (not just orchestrators)
+- Use tool call ordering (not agent narrative) to verify temporal constraints
+- MUST distinguish HARD-GATE violations (BLOCK) from constraint violations (WARN)
+- MUST NOT accept agent self-report — check independently via evidence
+
+Read `skills/constraint-check/SKILL.md` for the full specification.