feat(exodia): add Baryon Mode — one-shot .NET warning exterminator (#118)

* feat(exodia): add Baryon Mode — one-shot .NET warning exterminator

Noble Phantasm: snapshots tooling reality via dotnet build, then T0
burst launches up to 9 parallel agents (1 recon-invoker for MCP
version discovery + 8 domain aspects). No gates, no permission,
headless/reckless. Cross-repo with full MCP access.

Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>

* fix: address review findings — scope filtering, tools, counts, mktemp

1. Add WebFetch/WebSearch to allowed-tools (Codex)
2. Add $1 scope filtering logic — skip non-matching buckets (Codex P2)
3. Remove duplicate 'all' from scope options (Gemini)
4. Add CA2016 to async aspect in architecture diagram (Gemini)
5. Add / prefix to routing entry in CLAUDE.md (Codex)
6. Replace hardcoded /tmp paths with mktemp (Gemini)
7. Update README.md + AGENTS.md counts to 23/10 (Copilot)

Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>

---------

Co-authored-by: Claude Opus 4.6 <noreply@anthropic.com>

Author: ANcpLua

.claude-plugin/marketplace.json

@@ -5,7 +5,7 @@
     "url": "https://github.com/ANcpLua"
   },
   "metadata": {
-    "description": "Claude Code plugin marketplace: 7 plugins, 22 commands, 4 skills, 9 agents. Multi-agent orchestration, cognitive amplification, OpenTelemetry docs, and .NET build enforcement."
+    "description": "Claude Code plugin marketplace: 7 plugins, 23 commands, 4 skills, 9 agents. Multi-agent orchestration, cognitive amplification, OpenTelemetry docs, and .NET build enforcement."
   },
   "plugins": [
     {
@@ -46,7 +46,7 @@
     },
     {
       "name": "exodia",
-      "description": "Multi-agent workflow orchestration (9 commands + hades skill): fix, turbo-fix, fix-pipeline, tournament, mega-swarm, deep-think, batch-implement, red-blue-review, eight-gates.",
+      "description": "Multi-agent workflow orchestration (10 commands + hades skill): fix, turbo-fix, fix-pipeline, tournament, mega-swarm, deep-think, batch-implement, red-blue-review, eight-gates, baryon-mode.",
       "version": "2.0.0",
       "source": "./plugins/exodia"
     }

CHANGELOG.md

@@ -8,6 +8,7 @@ and the project follows [Semantic Versioning](https://semver.org/spec/v2.0.0.htm
 
 ### Added
 
+- **exodia `baryon-mode` command**: One-shot Noble Phantasm for .NET warning extermination. Phase 0 snapshots tooling reality via `dotnet build`, then T0 burst launches up to 9 parallel agents (1 recon-invoker for Rider MCP + NuGet MCP version discovery, 8 domain aspects: nullability, deprecation, unused code, async, style, suppressions, packages, config). No gates, no permission, headless/reckless. Cross-repo with full MCP access. Final verification via delta build comparison
 - **exodia `eight-gates` command**: Progressive discipline orchestration — 8 named gates (Kaimon→Shimon) composing scope, context loading (Yin), parallel MAP (Yang), checkpointing (Senzu), bounded reflection (Ralph Loop), reduce, TDD execution, and Hakai cleanup. Includes budget tracking, idempotent resume from any gate, TTL sessions, artifact caching, and decision logging. Composes mega-swarm (MAP), fix pipelines (EXECUTE), and hades (HAKAI) into a unified flow
 - **`checkpoint.sh` smart script**: Gate checkpoint management — init, save, load, verify (idempotent), list. Append-only JSONL storage with key=value metadata per gate
 - **`session-state.sh` smart script**: TTL session state + artifact cache + decision log. Create sessions with expiry, cache expensive computations, log decisions with reasons, extend/expire sessions

CLAUDE.md

@@ -122,8 +122,8 @@ ancplua-claude-plugins/
 │   ├── copilot-instructions.md
 │   └── workflows/               # CI, review, auto-merge
 │
-├── plugins/                     # 7 plugins (22 commands, 4 skills, 9 agents)
-│   ├── exodia/                  # Multi-agent orchestration — 9 commands + hades skill
+├── plugins/                     # 7 plugins (23 commands, 4 skills, 9 agents)
+│   ├── exodia/                  # Multi-agent orchestration — 10 commands + hades skill
 │   ├── metacognitive-guard/     # Cognitive amplification + commit integrity + CI
 │   ├── otelwiki/                # OpenTelemetry docs + sync
 │   ├── hookify/                 # User-configurable hooks
@@ -215,10 +215,14 @@ IF multi-agent orchestration needed
     /exodia:tournament        - competitive solutions (N+2 agents)
     /exodia:batch-implement   - parallel similar items (1+N+1 agents)
     /exodia:red-blue-review   - adversarial security (3+N+1 agents)
+    /exodia:baryon-mode       - .NET warning extermination (1+8 agents, one-shot T0 burst)
     /exodia:hades             - audited cleanup (3 phases x 4+3 teammates with goggles)
 
 IF zero-tolerance cleanup needed
   → exodia:hades skill (audited cleanup with Smart IDs, deletion permits, audit ledger)
+
+IF .NET warnings need extermination (one-shot, headless, cross-repo)
+  → /exodia:baryon-mode command (1 Invoker + 8 aspects burst at T0, full MCP access)
 ```
 
 ### Exodia Skills Detailed Routing
@@ -235,6 +239,7 @@ When multi-agent orchestration is needed, use these IF/THEN patterns:
 - **IF comprehensive codebase audit** THEN use `mega-swarm` skill (12 agents full, 6 quick, 8 focused)
 - **IF cleanup/elimination/dead code/suppressions** THEN use `hades` skill (Smart IDs, deletion permits, audit ledger)
 - **IF frontend cleanup + design quality** THEN use `hades --goggles` (Pink Glasses: taste → spec → compliance pipeline)
+- **IF .NET warning extermination** THEN use `baryon-mode` command (1 Invoker + 8 aspects, one-shot T0, cross-repo)
 
 ### Priority and Composition
 

README.md

@@ -2,13 +2,13 @@
 
 > **CCC** — Claude, Copilot, CodeRabbit. The holy trinity. And two of them are Claude in a trenchcoat.
 
-Claude Code plugin marketplace. 7 plugins, 22 commands, 4 skills, 9 agents, 18 scripts, 5 hook configs.
+Claude Code plugin marketplace. 7 plugins, 23 commands, 4 skills, 9 agents, 18 scripts, 5 hook configs.
 
 ## Plugins
 
 | Plugin | Version | Description |
 |--------|---------|-------------|
-| **exodia** | 2.0.0 | Multi-agent orchestration (9 commands + hades skill): fix, turbo-fix, fix-pipeline, tournament, mega-swarm, deep-think, batch-implement, red-blue-review, eight-gates |
+| **exodia** | 2.0.0 | Multi-agent orchestration (10 commands + hades skill): fix, turbo-fix, fix-pipeline, tournament, mega-swarm, deep-think, batch-implement, red-blue-review, eight-gates, baryon-mode |
 | **metacognitive-guard** | 0.4.0 | Cognitive amplification: epistemic hooks, commit integrity, CI verification, competitive review, deep analysis, deep-thinking agents |
 | **otelwiki** | 1.0.6 | OpenTelemetry docs with auto-sync and semantic convention validation |
 | **hookify** | 0.2.0 | User-configurable hooks from .local.md files |
@@ -31,7 +31,7 @@ Tri-AI review system: Claude, Copilot, CodeRabbit all review PRs independently.
 
 ```text
 plugins/
-├── exodia/                  # multi-agent orchestration (9 commands + hades cleanup skill)
+├── exodia/                  # multi-agent orchestration (10 commands + hades cleanup skill)
 ├── metacognitive-guard/     # struggle detection + deep-think + commit integrity + CI verification
 ├── otelwiki/                # OTel docs + sync
 ├── hookify/                 # user-configurable hooks

docs/AGENTS.md

@@ -23,4 +23,4 @@ Spawn agents via the Task tool with `subagent_type` matching the agent identifie
 The separation of concerns:
 - **Plugins** encode reusable behavior and skills
 - **Agents** orchestrate plugins and execute workflows
-- **Exodia commands** compose agents into multi-agent pipelines (9 commands + hades skill, 22 commands total across all plugins)
+- **Exodia commands** compose agents into multi-agent pipelines (10 commands + hades skill, 23 commands total across all plugins)

docs/ARCHITECTURE.md

@@ -35,8 +35,8 @@ ancplua-claude-plugins/
 │       ├── claude-code-review.yml # Claude formal PR review
 │       └── dependabot.yml       # Dependency update config
 │
-├── plugins/                     # 7 plugins (22 commands, 4 skills, 9 agents)
-│   ├── exodia/                  # Multi-agent orchestration (9 commands + hades skill)
+├── plugins/                     # 7 plugins (23 commands, 4 skills, 9 agents)
+│   ├── exodia/                  # Multi-agent orchestration (10 commands + hades skill)
 │   ├── metacognitive-guard/     # Cognitive amplification + commit integrity + CI
 │   ├── otelwiki/                # OpenTelemetry documentation + auto-sync
 │   ├── hookify/                 # User-configurable rule-based hooks
@@ -122,7 +122,7 @@ CI mirrors the same checks via `.github/workflows/ci.yml`.
 
 | Plugin | Responsibility |
 |--------|----------------|
-| `exodia` | Multi-agent orchestration (9 commands + hades skill) |
+| `exodia` | Multi-agent orchestration (10 commands + hades skill) |
 | `metacognitive-guard` | Cognitive amplification, commit integrity, CI verification |
 | `feature-dev` | Guided feature development + code review |
 | `otelwiki` | OpenTelemetry documentation + sync |

plugins/exodia/.claude-plugin/plugin.json

@@ -1,7 +1,7 @@
 {
   "name": "exodia",
   "version": "2.0.0",
-  "description": "Multi-agent workflow orchestration: fix, turbo-fix, fix-pipeline, tournament, mega-swarm, deep-think, batch-implement, red-blue-review, hades (Smart cleanup with audit trail).",
+  "description": "Multi-agent workflow orchestration (10 commands + hades skill): fix, turbo-fix, fix-pipeline, tournament, mega-swarm, deep-think, batch-implement, red-blue-review, eight-gates, baryon-mode.",
   "author": {
     "name": "AncpLua",
     "url": "https://github.com/ANcpLua"

plugins/exodia/CLAUDE.md

@@ -1,6 +1,6 @@
 # exodia
 
-Multi-agent workflow orchestration. 9 commands + 1 skill (hades), unlimited parallel agents.
+Multi-agent workflow orchestration. 10 commands + 1 skill (hades), unlimited parallel agents.
 
 ## Commands (in commands/)
 
@@ -15,6 +15,7 @@ Multi-agent workflow orchestration. 9 commands + 1 skill (hades), unlimited para
 | `deep-think` | 5 | Analysis before action, no implementation |
 | `batch-implement` | N+2 | Multiple similar items in parallel |
 | `red-blue-review` | 3+N+1 | Adversarial security/quality review |
+| `baryon-mode` | 1+8 | .NET warning extermination (one-shot T0 burst) |
 
 ## Skill (in skills/ — uses hooks/argument-hint)
 
@@ -45,4 +46,5 @@ Hookify guard templates in `scripts/smart/hookify-rules/`:
 - All commands/skills use blockquote teammate pattern (`> subagent: ... | model: ...`).
 - Descriptions encode IF/THEN routing (Vercel pattern).
 - Eight Gates composes all others: mega-swarm (MAP) -> fix pipelines (EXECUTE) -> hades (HAKAI).
+- Baryon Mode is a Noble Phantasm: one-shot T0 burst for .NET warnings. No gates, no permission. Invoker + 8 aspects.
 - Exodia creates, Hades judges. They compose: mega-swarm -> hades.

plugins/exodia/README.md

@@ -1,6 +1,6 @@
 # exodia v2.0.0
 
-Multi-agent workflow orchestration. 9 commands + 1 skill (hades).
+Multi-agent workflow orchestration. 10 commands + 1 skill (hades).
 
 ## Commands
 
@@ -15,6 +15,7 @@ Multi-agent workflow orchestration. 9 commands + 1 skill (hades).
 | `/exodia:deep-think` | 5 | Multi-perspective | Analysis before action |
 | `/exodia:batch-implement` | N+2 | Template + parallel | Similar items |
 | `/exodia:red-blue-review` | 3+N | Adversarial | Security review |
+| `/exodia:baryon-mode` | 1+8 | One-shot T0 burst | .NET warning extermination |
 
 ## Skill (uses hooks/argument-hint)
 
@@ -27,11 +28,12 @@ Multi-agent workflow orchestration. 9 commands + 1 skill (hades).
 ```text
 1. Go all-in on objective     → /exodia:eight-gates "ship v2.0" . 8
 2. Audit the codebase         → /exodia:mega-swarm mode=quick
-3. Fix critical issues        → /exodia:turbo-fix (P0) or /exodia:fix (P1/P2)
-4. Security review            → /exodia:red-blue-review scope=security
-5. Architectural decisions    → /exodia:deep-think mode=architecture
-6. Batch implement features   → /exodia:batch-implement type=endpoints
-7. Re-audit                   → /exodia:mega-swarm mode=full
+3. Exterminate warnings       → /exodia:baryon-mode path/to/Solution.sln
+4. Fix critical issues        → /exodia:turbo-fix (P0) or /exodia:fix (P1/P2)
+5. Security review            → /exodia:red-blue-review scope=security
+6. Architectural decisions    → /exodia:deep-think mode=architecture
+7. Batch implement features   → /exodia:batch-implement type=endpoints
+8. Re-audit                   → /exodia:mega-swarm mode=full
 ```
 
 ## Installation