@trac3er/oh-my-god 2.0.4 → 2.0.7

package/docs/migration/native-adoption.md

@@ -0,0 +1,57 @@
+# Native Adoption for OMC, OMX, and Superpowers
+
+OMG keeps adoption native to OMG. There is no public `migrate` command to memorize or map back to another project.
+
+Use either:
+
+- `/OMG:setup`
+- `./OMG-setup.sh install --adopt=auto`
+
+## What Setup Detects
+
+Setup can detect compatibility markers from:
+
+- OMC-style environments
+- OMX-style environments
+- Superpowers-style command, skill, or workspace surfaces
+
+Those detections are treated as adoption context, not as public implementation detail.
+
+## Adoption Modes
+
+### OMG-only
+
+Recommended for most users.
+
+- OMG becomes the primary hooks, HUD, MCP, and orchestration layer.
+- Portable state is preserved where safe and overlapping surfaces are backed up before replacement.
+- `compat` remains available for legacy skill routing only.
+
+### coexist
+
+Advanced mode for careful landings.
+
+- OMG avoids claiming ownership of third-party namespaces.
+- Existing HUD and hook ownership stays intact unless you explicitly switch later.
+- Conflicts are recorded in the adoption report instead of being overwritten.
+
+## Output
+
+Native adoption writes:
+
+- `.omg/state/adoption-report.json`
+- `.omg/state/cli-config.yaml`
+- `.mcp.json`
+
+## Choosing a Preset
+
+- `safe`: conservative OMG baseline
+- `balanced`: recommended daily workflow
+- `interop`: stronger shared-memory and multi-host interop
+- `labs`: enables the most opinionated experimentation surfaces
+
+## What Changes for Existing Users
+
+- OMG-native setup language replaces public migration language.
+- Compatibility references remain documented so users coming from OMC, OMX, or Superpowers can understand what setup detected.
+- Public onboarding stays focused on `/OMG:setup` and `/OMG:crazy`.

package/docs/proof.md

@@ -0,0 +1,65 @@
+# OMG Proof Surface
+
+[![Compat Gate](https://github.com/trac3er00/OMG/actions/workflows/omg-compat-gate.yml/badge.svg)](https://github.com/trac3er00/OMG/actions/workflows/omg-compat-gate.yml)
+[![npm version](https://img.shields.io/npm/v/%40trac3er%2Foh-my-god)](https://www.npmjs.com/package/@trac3er/oh-my-god)
+
+## Verification Status
+
+OMG keeps verification visible instead of burying it in implementation details.
+
+- Runtime evidence root: `.omg/evidence/`
+- Doctor output: `.omg/evidence/doctor.json`
+- Security-check artifacts: `.omg/evidence/security-check-*.json`
+- Trust and external input artifacts: `.omg/evidence/trust-*.json`
+- Truth bundles:
+  - `claim-judge`: `.omg/evidence/claim-judge-*.json` (verifies claim-to-evidence mapping)
+  - `test-intent-lock`: `.omg/evidence/test-intent-lock-*.json` (verifies test-to-intent alignment)
+  - `proof-gate`: `.omg/evidence/proof-gate-*.json` (verifies final release readiness)
+- Browser evidence: `.omg/evidence/browser-*.png` and `.omg/evidence/browser-*.json` (Playwright-backed verification)
+- Trace records and evidence links: `.omg/tracebank/events.jsonl`, `.omg/tracebank/evidence-links.jsonl`
+- Eval gate artifacts and trace links: `.omg/evals/latest.json`, `.omg/evals/history.jsonl`, `.omg/evals/trace-links.jsonl`
+- Lineage manifests: `.omg/lineage/*.json`
+- Release readiness output links these machine artifacts instead of prose-only pass counts.
+
+## Provider Matrix
+
+| Provider | Detect | Auth Check | MCP Config | Host Priority |
+|----------|--------|------------|------------|---------------|
+| Claude Code | host-native | host-native | yes | primary |
+| Codex | yes | yes | yes | primary |
+| Gemini | yes | yes | yes | supported |
+| Kimi | yes | yes | yes | supported |
+
+## Adoption Evidence
+
+- Native setup writes `.omg/state/adoption-report.json`
+- Native setup writes `.omg/state/cli-config.yaml`
+- `OMG-only` and `coexist` are both covered in setup tests
+- OMC, OMX, and Superpowers references stay limited to compatibility and adoption guidance
+
+## HUD Artifact
+
+![OMG HUD](assets/omg-hud.svg)
+
+## Benchmark Tasks
+
+Representative benchmark tasks for this release:
+
+- host detection and auth wiring
+- canonical security-check routing and evidence emission
+- narrowed stdio OMG control MCP wiring
+- truth bundle verification (claim-judge, test-intent-lock, proof-gate)
+- plan-council role compilation and execution
+- adoption detection with overlapping ecosystems
+- plugin install and uninstall correctness
+- `crazy` orchestration smoke coverage
+
+## Sample Transcripts
+
+- Setup: [docs/transcripts/setup.md](transcripts/setup.md)
+- Crazy: [docs/transcripts/crazy.md](transcripts/crazy.md)
+
+## Release Discipline
+
+- Public launch checklist: [docs/release-checklist.md](release-checklist.md)
+- Changelog: [CHANGELOG.md](../CHANGELOG.md)

package/docs/release-checklist.md

@@ -0,0 +1,40 @@
+# Public Release Checklist
+
+Use this checklist before making OMG public or cutting a release tag.
+
+## Identity
+
+- README, `package.json`, plugin manifests, and CLI version output agree on the version
+- repo URL is `https://github.com/trac3er00/OMG`
+- plugin and marketplace id are `omg`
+
+## Public Safety
+
+- `python3 scripts/check-omg-public-ready.py` passes
+- no internal planning docs ship in `docs/plans/`
+- no absolute local paths or stale internal path references remain
+
+## Verification
+
+- `python3 scripts/omg.py contract validate` passes
+- `python3 scripts/omg.py contract compile --host claude --host codex --channel public --output-root <tmp>` passes
+- `python3 scripts/omg.py contract compile --host claude --host codex --channel enterprise --output-root <tmp>` passes
+- `python3 scripts/omg.py release readiness --channel dual --output-root <tmp>` passes
+- Truth bundles (`claim-judge`, `test-intent-lock`, `proof-gate`) are verified and present in `registry/bundles/`
+- `plan-council` role is compiled and verified in `registry/bundles/`
+- `python3 scripts/check-omg-standalone-clean.py` passes
+- `./scripts/verify-standalone.sh` passes
+- `python3 -m pytest tests -q` passes
+
+## Docs
+
+- README matches the current product surface
+- install guides for Claude Code and Codex are current
+- proof page includes current verification evidence
+- changelog includes the release entry
+
+## Release Ops
+
+- `package.json` version matches the tag you will create
+- npm publishing credentials are configured for `publish-npm.yml`
+- GitHub Actions required checks are green

package/docs/transcripts/crazy.md

@@ -0,0 +1,17 @@
+# Sample Transcript: /OMG:crazy
+
+```text
+/OMG:crazy stabilize auth and dashboard flows
+
+Workers:
+- architect-mode
+- backend-engineer
+- frontend-designer
+- security-auditor
+- testing-engineer
+
+Execution:
+- parallel
+- evidence-aware
+- verification required before completion
+```

package/docs/transcripts/setup.md

@@ -0,0 +1,25 @@
+# Sample Transcript: /OMG:setup
+
+```text
+/OMG:setup
+
+Detected hosts:
+- codex: installed, authenticated
+- gemini: not found
+- kimi: not found
+
+Detected ecosystems:
+- omc
+- superpowers
+
+Recommended mode: omg-only
+Selected preset: balanced
+
+Configured:
+- .mcp.json
+- .omg/state/cli-config.yaml
+- .omg/state/adoption-report.json
+
+Next step:
+/OMG:crazy stabilize auth and dashboard flows
+```

package/hooks/policy_engine.py

@@ -8,6 +8,7 @@ from __future__ import annotations
 
 from dataclasses import dataclass, asdict
 from fnmatch import fnmatch
+import importlib
 import os
 import re
 from typing import Any
@@ -122,12 +123,89 @@ UNTRUSTED_MUTATION_PATTERNS = [
     r"\b(mv|cp|tee|sed\s+-i|touch|mkdir)\b",
 ]
 
+TRUSTED_CONTENT_TIERS = frozenset({"local", "balanced"})
+UNTRUSTED_EXTERNAL_TIERS = frozenset({"research", "browser"})
+
+
+def _project_dir() -> str:
+    return os.environ.get("CLAUDE_PROJECT_DIR", os.getcwd())
+
+
+def _load_untrusted_provenance_entries() -> list[dict[str, Any]]:
+    try:
+        from runtime.untrusted_content import get_untrusted_content_state
+
+        state = get_untrusted_content_state(_project_dir())
+        provenance = state.get("provenance", [])
+        if isinstance(provenance, list):
+            return [entry for entry in provenance if isinstance(entry, dict)]
+    except Exception:
+        return []
+    return []
+
+
+def _is_state_changing_action(action: str) -> bool:
+    normalized = str(action).strip().lower()
+    return normalized in {
+        "state_change",
+        "state-changing",
+        "bash_mutation",
+        "file_mutation",
+        "write",
+        "edit",
+        "delete",
+    }
+
+
+def evaluate_action_justification(
+    *,
+    action: str,
+    evidence: list[dict[str, Any]],
+    require_explicit_approval: bool = True,
+) -> PolicyDecision:
+    if not _is_state_changing_action(action):
+        return allow("non-mutating action")
+    if not evidence:
+        return ask(
+            "State-changing action lacks trust-scored evidence.",
+            "high",
+            ["manual-approval", "trusted-evidence-required"],
+        )
+
+    tiers = {
+        str(item.get("_trust_tier") or item.get("trust_tier") or "").strip().lower()
+        for item in evidence
+        if isinstance(item, dict)
+    }
+    tiers.discard("")
+    has_trusted = bool(tiers & TRUSTED_CONTENT_TIERS)
+    has_external_only = bool(tiers) and tiers.issubset(UNTRUSTED_EXTERNAL_TIERS)
+
+    if has_trusted:
+        return allow("trusted local evidence present")
+
+    if has_external_only:
+        reason = (
+            "State-changing action is justified only by UNTRUSTED_EXTERNAL_CONTENT "
+            "(research/browser tier)."
+        )
+        controls = ["manual-approval", "trusted-corroboration", "review-provenance"]
+        if require_explicit_approval:
+            return ask(reason, "high", controls)
+        return deny(reason, "high", controls)
+
+    return ask(
+        "State-changing action has unknown trust provenance.",
+        "high",
+        ["manual-approval", "review-provenance"],
+    )
+
 
 def _is_untrusted_content_mode_active() -> bool:
     try:
         from runtime.untrusted_content import is_untrusted_content_mode_active
 
-        project_dir = os.environ.get("CLAUDE_PROJECT_DIR", os.getcwd())
+        project_dir = _project_dir()
         return is_untrusted_content_mode_active(project_dir)
     except Exception:
         return False
@@ -175,14 +253,25 @@ def evaluate_bash_command(cmd: str) -> PolicyDecision:
         if re.search(pat, cmd):
             return ask(f"{label}: {cmd[:120]}", "med", ["human-approval"])
 
-    if _is_untrusted_content_mode_active():
-        for pat in UNTRUSTED_MUTATION_PATTERNS:
-            if re.search(pat, cmd):
-                return ask(
-                    "Untrusted external content mode is active. Review before running state-changing commands.",
-                    "high",
-                    ["manual-approval", "review-provenance"],
-                )
+    for pat in UNTRUSTED_MUTATION_PATTERNS:
+        if not re.search(pat, cmd):
+            continue
+        provenance_entries = _load_untrusted_provenance_entries()
+        if provenance_entries:
+            decision = evaluate_action_justification(
+                action="state_change",
+                evidence=provenance_entries,
+                require_explicit_approval=True,
+            )
+            if decision.action != "allow":
+                return decision
+        if _is_untrusted_content_mode_active():
+            return ask(
+                "Untrusted external content mode is active. Review before running state-changing commands.",
+                "high",
+                ["manual-approval", "review-provenance"],
+            )
+        break
 
     return allow("command allowed")
 
@@ -232,7 +321,10 @@ def _is_omg_credential_path(normalized_path: str) -> bool:
     This is deliberately narrow to prevent path traversal attacks.
     """
     # Import here to avoid circular dependency at module level
-    from _common import get_feature_flag
+    try:
+        get_feature_flag = getattr(importlib.import_module("hooks._common"), "get_feature_flag")
+    except Exception:
+        get_feature_flag = getattr(importlib.import_module("_common"), "get_feature_flag")
 
     # Only exempt if feature is enabled
     if not get_feature_flag("MULTI_CREDENTIAL", default=False):
@@ -413,7 +505,10 @@ def _log_allowlist_bypass(path: str, tool: str, reason: str) -> None:
     Silently fails — never raises exceptions (crash isolation invariant).
     """
     try:
-        from secret_audit import log_secret_access
+        try:
+            log_secret_access = getattr(importlib.import_module("hooks.secret_audit"), "log_secret_access")
+        except Exception:
+            log_secret_access = getattr(importlib.import_module("secret_audit"), "log_secret_access")
 
         project_dir = os.environ.get("CLAUDE_PROJECT_DIR", os.getcwd())
         log_secret_access(
@@ -472,12 +567,22 @@ def evaluate_file_access(
         if re.search(pat, lowpath):
             return deny(f"Sensitive path blocked: {file_path}", "critical", ["secret-access"])
 
-    if tool in {"Write", "Edit", "MultiEdit"} and _is_untrusted_content_mode_active():
-        return ask(
-            "Untrusted external content mode is active. Review before mutating files.",
-            "high",
-            ["manual-approval", "review-provenance"],
-        )
+    if tool in {"Write", "Edit", "MultiEdit"}:
+        provenance_entries = _load_untrusted_provenance_entries()
+        if provenance_entries:
+            decision = evaluate_action_justification(
+                action="file_mutation",
+                evidence=provenance_entries,
+                require_explicit_approval=True,
+            )
+            if decision.action != "allow":
+                return decision
+        if _is_untrusted_content_mode_active():
+            return ask(
+                "Untrusted external content mode is active. Review before mutating files.",
+                "high",
+                ["manual-approval", "review-provenance"],
+            )
 
     if allowlist and is_allowlisted(file_path, tool, allowlist):
         return allow(f"Allowlisted: {file_path}")

package/hooks/setup_wizard.py

@@ -69,6 +69,9 @@ BYPASS_ALL_WARNING = (
     "Do you want to enable bypass-all mode? (y/N): "
 )
 
+PRESET_ORDER: tuple[str, ...] = ("safe", "balanced", "interop", "labs")
+_PRESET_LEVEL: dict[str, int] = {p: i for i, p in enumerate(PRESET_ORDER)}
+
 MCP_CATALOG: list[dict[str, Any]] = [
     {
         "id": "context7",
@@ -76,7 +79,8 @@ MCP_CATALOG: list[dict[str, Any]] = [
         "description": "Upstash Context7 MCP server for context management",
         "command": "npx",
         "args": ["@upstash/context7-mcp@2.1.3"],
-        "default": True,
+        "default": False,
+        "min_preset": "balanced",
         "category": "productivity",
     },
     {
@@ -86,6 +90,7 @@ MCP_CATALOG: list[dict[str, Any]] = [
         "command": "npx",
         "args": ["@modelcontextprotocol/server-filesystem@2026.1.14", "."],
         "default": True,
+        "min_preset": "safe",
         "category": "system",
     },
     {
@@ -94,7 +99,8 @@ MCP_CATALOG: list[dict[str, Any]] = [
         "description": "Web search MCP server for internet queries",
         "command": "npx",
         "args": ["@zhafron/mcp-web-search@1.2.2"],
-        "default": True,
+        "default": False,
+        "min_preset": "interop",
         "category": "search",
     },
     {
@@ -103,7 +109,8 @@ MCP_CATALOG: list[dict[str, Any]] = [
         "description": "Chrome DevTools MCP server for browser automation",
         "command": "npx",
         "args": ["chrome-devtools-mcp@0.19.0"],
-        "default": True,
+        "default": False,
+        "min_preset": "labs",
         "category": "browser",
     },
     {
@@ -114,7 +121,8 @@ MCP_CATALOG: list[dict[str, Any]] = [
         "args": [],
         "type": "http",
         "url": "http://127.0.0.1:8765/mcp",
-        "default": True,
+        "default": False,
+        "min_preset": "interop",
         "category": "memory",
     },
     {
@@ -124,6 +132,7 @@ MCP_CATALOG: list[dict[str, Any]] = [
         "command": OMG_CONTROL_COMMAND,
         "args": OMG_CONTROL_ARGS,
         "default": True,
+        "min_preset": "safe",
         "category": "control",
     },
     {
@@ -192,6 +201,26 @@ def get_mcp_catalog() -> list[dict[str, Any]]:
     return MCP_CATALOG
 
 
+def get_default_mcps_for_preset(preset: str) -> list[str]:
+    """Return the list of MCP server IDs enabled by default for *preset*.
+
+    Each catalog entry carries a ``min_preset`` field that specifies the
+    lowest preset tier at which the MCP is included.  The preset order is
+    ``safe < balanced < interop < labs``.
+
+    Returns:
+        Sorted list of MCP server IDs whose ``min_preset`` is at or below
+        the requested *preset* level.
+    """
+    level = _PRESET_LEVEL.get(preset, 0)
+    return [
+        m["id"]
+        for m in MCP_CATALOG
+        if _PRESET_LEVEL.get(m.get("min_preset", ""), -1) <= level
+        and m.get("min_preset") is not None
+    ]
+
+
 def build_mcp_config(selected_ids: list[str]) -> dict[str, Any]:
     """Build .mcp.json configuration from selected MCP server IDs.
 
@@ -245,17 +274,18 @@ def configure_plan_type(plan_type: str) -> dict[str, Any]:
     return result
 
 
-def select_mcps(selected_ids: list[str] | None = None) -> dict[str, Any]:
+def select_mcps(selected_ids: list[str] | None = None, preset: str = "safe") -> dict[str, Any]:
     """Build MCP config from selected IDs.
 
     Args:
-        selected_ids: List of MCP IDs to include. If None, uses defaults.
+        selected_ids: List of MCP IDs to include. If None, uses preset defaults.
+        preset: Preset tier that controls which MCPs are included by default.
 
     Returns:
         dict with mcpServers key (ready to write as .mcp.json)
     """
     if selected_ids is None:
-        selected_ids = [m["id"] for m in MCP_CATALOG if m["default"]]
+        selected_ids = get_default_mcps_for_preset(preset)
     return build_mcp_config(selected_ids)
 
 
@@ -383,6 +413,9 @@ def check_auth() -> dict[str, Any]:
     return {"status": "pending", "results": {}}
 
 
+_HTTP_MEMORY_MIN_LEVEL: int = _PRESET_LEVEL["interop"]
+
+
 def configure_mcp(
     project_dir: str,
     detected_clis: dict[str, Any],
@@ -391,12 +424,17 @@ def configure_mcp(
     control_command: str = OMG_CONTROL_COMMAND,
     control_args: list[str] | None = None,
     control_server_name: str = OMG_CONTROL_SERVER_NAME,
+    preset: str = "safe",
 ) -> dict[str, Any]:
     """Configure OMG MCP servers for authenticated CLIs.
 
     For each CLI in detected_clis where detected_clis[cli]["detected"] == True,
     calls the appropriate writer from runtime.mcp_config_writers.
 
+    HTTP memory surfaces are only written when the *preset* is at or above
+    ``interop`` level.  ``safe`` and ``balanced`` presets write only the
+    stdio ``omg-control`` surface.
+
     Args:
         project_dir: Path to the project directory.
         detected_clis: Dict of CLI detection results from detect_clis().
@@ -405,6 +443,7 @@ def configure_mcp(
         control_command: stdio command for the OMG control MCP server.
         control_args: stdio args for the OMG control MCP server.
         control_server_name: MCP server name for the OMG control surface.
+        preset: Active preset tier (safe, balanced, interop, labs).
 
     Returns:
         Dict with keys:
@@ -415,10 +454,11 @@ def configure_mcp(
     configured: list[str] = []
     errors: dict[str, str] = {}
     resolved_control_args = list(control_args or OMG_CONTROL_ARGS)
+    http_memory_allowed = _PRESET_LEVEL.get(preset, 0) >= _HTTP_MEMORY_MIN_LEVEL
 
-    # Always write Claude project config for both memory and control surfaces.
     try:
-        write_claude_mcp_config(project_dir, server_url, server_name)
+        if http_memory_allowed:
+            write_claude_mcp_config(project_dir, server_url, server_name)
         write_claude_mcp_stdio_config(
             project_dir,
             command=control_command,
@@ -429,7 +469,6 @@ def configure_mcp(
         _logger.warning("Failed to write Claude MCP config: %s", exc)
         errors["claude"] = str(exc)
 
-    # Write both memory and control surfaces for detected external CLIs.
     cli_writers = {
         "codex": (write_codex_mcp_config, write_codex_mcp_stdio_config),
         "gemini": (write_gemini_mcp_config, write_gemini_mcp_stdio_config),
@@ -442,7 +481,8 @@ def configure_mcp(
             continue
 
         try:
-            http_writer(server_url, server_name)
+            if http_memory_allowed:
+                http_writer(server_url, server_name)
             stdio_writer(
                 command=control_command,
                 args=resolved_control_args,
@@ -591,7 +631,7 @@ def run_setup_wizard(
 
     clis = detect_clis()
     auth = check_auth()
-    mcp = configure_mcp(project_dir, clis)
+    mcp = configure_mcp(project_dir, clis, preset=selected_preset)
     prefs = set_preferences(project_dir, {"preset": selected_preset})
     report_path = write_adoption_report(project_dir, adoption)
     adoption["report_path"] = report_path

package/hooks/shadow_manager.py

@@ -8,7 +8,9 @@ from __future__ import annotations
 import hashlib
 import json
 import os
+import platform
 import shutil
+import socket
 import sys
 from datetime import datetime, timezone
 from typing import Any
@@ -182,9 +184,29 @@ def create_evidence_pack(
     provenance: list[dict[str, Any]] | None = None,
     trust_scores: dict[str, Any] | None = None,
     api_twin: dict[str, Any] | None = None,
+    route_metadata: dict[str, Any] | None = None,
+    trace_ids: list[str] | None = None,
+    lineage: dict[str, Any] | None = None,
+    executor: dict[str, Any] | None = None,
+    environment: dict[str, Any] | None = None,
 ) -> str:
     ensure_shadow_dirs(project_dir)
     run_id = _validated_run_id(run_id)
+    
+    # Default executor if not provided
+    if executor is None:
+        executor = {
+            "user": os.getenv("USER", "unknown"),
+            "pid": str(os.getpid()),
+        }
+    
+    # Default environment if not provided
+    if environment is None:
+        environment = {
+            "hostname": socket.gethostname(),
+            "platform": platform.system(),
+        }
+    
     evidence = {
         "schema": "EvidencePack",
         "run_id": run_id,
@@ -197,6 +219,11 @@ def create_evidence_pack(
         "provenance": provenance or [],
         "trust_scores": trust_scores or {},
         "api_twin": api_twin or {},
+        "route_metadata": route_metadata or {},
+        "trace_ids": trace_ids or [],
+        "lineage": lineage or {},
+        "executor": executor,
+        "environment": environment,
     }
     evidence_path = ensure_path_within_dir(
         _evidence_root(project_dir),

package/package.json

@@ -1,7 +1,7 @@
 {
   "name": "@trac3er/oh-my-god",
-  "version": "2.0.4",
-  "description": "OMG (Oh My God) — Multi-agent orchestration, intelligent model routing, and durable session state for Claude Code",
+  "version": "2.0.7",
+  "description": "OMG (Oh My God) — Multi-agent orchestration, evidence-backed verification, and durable session state for Claude Code, Codex, and supported agent hosts",
   "main": "OMG-setup.sh",
   "scripts": {
     "postinstall": "./OMG-setup.sh install --non-interactive",

package/plugins/README.md

@@ -28,11 +28,15 @@ OMG exposes a small native front door and keeps the rest of the surface availabl
 
 | Command | Category | Description |
 |---------|----------|-------------|
-| `/OMG:deep-plan` | Planning | Strategic planning with domain awareness |
+| `/OMG:deep-plan` | Planning | Strategic planning with domain awareness (routes to `plan-council`) |
 | `/OMG:learn` | Knowledge | Convert patterns into OMG-native instincts and skills |
 | `/OMG:code-review` | Quality | Deep review flow |
 | `/OMG:ship` | Delivery | Idea to evidence to release |
 | `/OMG:handoff` | Collaboration | Session transfer and continuity |
+| `/OMG:maintainer` | OSS | Open-source maintainer workflows |
+| `/OMG:sequential-thinking` | Thinking | Structured multi-step reasoning |
+| `/OMG:ralph-start` | Automation | Start Ralph autonomous loop |
+| `/OMG:ralph-stop` | Automation | Stop Ralph autonomous loop |
 
 ## Plugin Layout
 

package/plugins/__init__.py

@@ -0,0 +1 @@
+"""OMG plugin packages."""