aiwg 2026.9.2 → 2026.9.4

package/README.md

@@ -1,10 +1,10 @@
 <div align="center">
 
-<a href="https://aiwg.io"><img src="https://aiwg.io/assets/badges/aiwg-hero-dark.png" alt="AIWG — multi-agent AI framework · one source of truth · 12 provider integrations" width="680"></a>
+<a href="https://aiwg.io"><img src="https://aiwg.io/assets/badges/aiwg-hero-dark.png" alt="AIWG — multi-agent AI framework · one source of truth · 14 provider integrations including Google Antigravity CLI" width="680"></a>
 
 # AIWG
 
-**Multi-agent AI framework for 12 provider integrations, including Claude Code, Codex, Copilot, Cursor, Pi, and Warp**
+**Multi-agent AI framework for 14 provider integrations, including Antigravity, Claude Code, Codex, Copilot, Cursor, Pi, and Warp**
 
 200+ agents, 109+ CLI commands, 400+ deployable agent/skill/command/rule artifacts, 8 core frameworks, 32 addons, and a 40-plugin Claude Code marketplace. SDLC workflows, digital forensics, research management, marketing operations, media curation, ops infrastructure, knowledge base, and fine-tuning dataset curation — all deployable with one command.
 
@@ -47,7 +47,7 @@ aiwg use all --provider <provider>
 ```
 
 Replace `<provider>` with your AI tool's name, such as `claude`, `codex`,
-`copilot`, or `cursor`.
+`copilot`, `cursor`, `omp`, or `pi`.
 
 The final command deploys, indexes, connects, verifies, and reports one outcome.
 The standalone index, regenerate, status, and doctor commands remain available
@@ -85,7 +85,7 @@ Agents and stewards setting up AIWG end-to-end should use the
 [![GitHub Stars](https://img.shields.io/github/stars/jmagly/aiwg?style=flat-square)](https://github.com/jmagly/aiwg/stargazers)
 [![Node Version](https://img.shields.io/badge/node-%E2%89%A520.0.0-brightgreen?style=flat-square&logo=node.js)](https://nodejs.org)
 [![TypeScript](https://img.shields.io/badge/TypeScript-5.x-blue?style=flat-square&logo=typescript)](https://www.typescriptlang.org)
-[![12 Providers](https://img.shields.io/badge/Providers-12-purple?style=flat-square)](#-platform-support)
+[![14 Providers](https://img.shields.io/badge/Providers-14-purple?style=flat-square)](#-platform-support)
 [![Listed on mcpservers.org](https://mcpservers.org/badge.svg)](https://mcpservers.org/servers/docs-aiwg-io)
 
 [![Built With AIWG](https://aiwg.io/assets/badges/built-with-aiwg-dark.png)](https://aiwg.io/badges)
@@ -164,7 +164,7 @@ You can also invoke AIWG without adjusting `PATH` by using `npx aiwg <command>`.
 
 ## What AIWG Is
 
-AIWG is a deployment tool and support utility for AI context. At its core, `aiwg use` copies markdown and YAML source files into the paths each provider reads, so one source of truth works across 12 named provider integrations. A thirteenth `generic` adapter emits portable files for unrecognized or custom harnesses and is not counted as a named integration.
+AIWG is a deployment tool and support utility for AI context. At its core, `aiwg use` copies markdown and YAML source files into the paths each provider reads, so one source of truth works across 14 named provider integrations. A fifteenth `generic` adapter emits portable files for unrecognized or custom harnesses and is not counted as a named integration.
 
 Around that core, AIWG ships agent-facing utilities for things the base platforms do not handle on their own: persistent artifact memory (`.aiwg/`), background orchestration, autonomous loops, artifact indexing, cost telemetry, health diagnostics, and more. These are tools the agent calls when you ask for something AIWG-shaped — you stay in chat. Most are opt-in. The deployment layer works standalone as plain text files the platform reads natively.
 
@@ -543,7 +543,7 @@ The orchestration pattern: **Primary Author → Parallel Reviewers → Synthesiz
 - **128 workflow skills** — natural language triggers for regression testing, forensics, voice profiles, quality gates, and CI/CD integration
 - **35 enforcement rules** — anti-laziness detection, token security, citation integrity, executable feedback, failure mitigation across 6 LLM archetypes
 - **334 artifact templates** — progressive disclosure templates for requirements, architecture, testing, security, deployment, and more
-- **Multi-provider support** — deploy to Claude Code, OpenAI Codex, GitHub Copilot, Cursor, Factory AI, Hermes, OpenCode, OpenClaw, OpenHuman, Pi Coding Agent, Warp Terminal, and Devin Desktop
+- **Multi-provider support** — deploy to Google Antigravity CLI, Claude Code, OpenAI Codex, GitHub Copilot, Cursor, Factory AI, Hermes, OpenCode, OpenClaw, OpenHuman, [Pi Coding Agent](https://pi.dev/), Oh My Pi, Warp Terminal, and Devin Desktop
 - **8 core frameworks + training marketplace package** — SDLC, Digital Forensics, Marketing Operations, Research Management, Media Curation, Ops Infrastructure, Knowledge Base, Security Engineering, plus [`aiwg-training`](https://github.com/jmagly/aiwg-training) for fine-tuning dataset curation (corpus-to-dataset pipeline with DPO/KTO/ORPO/SimPO export)
 - **32 addons** — compound memory, line memory, llm-wiki (Obsidian-native knowledge base), RLM recursive decomposition, fleet operations, browser control, testing quality, and more
 - **40 Claude Code plugins** — the complete framework and addon catalog is installable independently from the AIWG marketplace
@@ -645,6 +645,7 @@ needs. Source-distributed opt-in addons such as Civic Action deploy with
 ### Multi-Platform Deployment
 
 ```bash
+aiwg use all --provider antigravity    # Google Antigravity CLI (alias: agy)
 aiwg use all --provider claude         # Claude Code
 aiwg use all --provider codex          # OpenAI Codex
 aiwg use all --provider copilot        # GitHub Copilot
@@ -1384,10 +1385,11 @@ See [Messaging Guide](docs/messaging-guide.md) for setup and configuration.
 
 ## Platform Support
 
-AIWG supports 12 named provider integrations. Artifact support varies by provider and is adapted to each provider's native or compatibility surfaces.
+AIWG supports 14 named provider integrations. Artifact support varies by provider and is adapted to each provider's native or compatibility surfaces.
 
 | Platform | Status | Agents | Commands | Skills | Rules | Deploy Command |
 |----------|--------|--------|----------|--------|-------|---------------|
+| **[Google Antigravity CLI](docs/providers/antigravity.md)** | Experimental | degraded `.agents/agents/` | — | `.agents/skills/` | `AGENTS.md` | `--provider antigravity` (alias: `agy`) |
 | **Claude Code** | Tested | `.claude/agents/` | `.claude/commands/` | `.claude/skills/` | `.claude/rules/` | `aiwg use sdlc` |
 | **GitHub Copilot** | Tested | `.github/agents/` | `.github/agents/` | `.github/skills/` | `.github/copilot-rules/` | `--provider copilot` |
 | **Warp Terminal** | Tested | `.warp/agents/` + WARP.md | `.warp/commands/` | `.warp/skills/` | `.warp/rules/` | `--provider warp` |
@@ -1396,10 +1398,11 @@ AIWG supports 12 named provider integrations. Artifact support varies by provide
 | **OpenCode** | Tested | `.opencode/agent/` | `.opencode/commands/` | `.opencode/skill/` | `.opencode/rule/` | `--provider opencode` |
 | **OpenAI/Codex** | Tested | `.codex/agents/` | `~/.codex/prompts/` | `.agents/skills/` | `.codex/rules/` | `--provider codex` |
 | **Devin Desktop** | Tested compatibility adapter | AGENTS.md | `.windsurf/workflows/` | `.windsurf/skills/` | `.windsurf/rules/` | `--provider devin` |
-| **Hermes** | Experimental | — | — | `~/.hermes/skills/.aiwg/` | — | `--provider hermes` |
+| **Hermes** | Stable | — | — | `~/.hermes/skills/.aiwg/` | — | `--provider hermes` |
 | **OpenClaw** | Tested | `~/.openclaw/agents/` | `~/.openclaw/commands/` | `~/.openclaw/.aiwg/skills/` | `~/.openclaw/rules/` | `--provider openclaw` |
 | **OpenHuman** | Experimental | — | — | `~/.openhuman/.aiwg/skills/` | `~/.openhuman/.aiwg/rules/` | `--provider openhuman` |
-| **Pi Coding Agent** | Experimental | `.agents/skills/` | `.pi/prompts/` | `.agents/skills/` | `AGENTS.md` | `--provider pi` |
+| **[Pi Coding Agent](https://pi.dev/)** | Experimental | `.agents/skills/` | `.pi/prompts/` | `.agents/skills/` | `AGENTS.md` | `--provider pi` |
+| **[Oh My Pi](docs/providers/omp.md)** | Experimental | `.omp/agents/` | `.omp/prompts/` | `.agents/skills/` | `.omp/AGENTS.md` | `--provider omp` |
 
 The legacy `--provider windsurf` selector remains supported and writes the
 same `.windsurf/` compatibility paths, but new commands should use `devin`.
@@ -1479,7 +1482,7 @@ aiwg sdlc-accelerate --from-codebase .
 
 ### Extension System
 
-AIWG uses a unified extension system with 10 extension types, projected onto the supported artifact surfaces of 12 named provider integrations:
+AIWG uses a unified extension system with 10 extension types, projected onto the supported artifact surfaces of 14 named provider integrations:
 
 | Type | Count | Description |
 |------|-------|-------------|
@@ -1857,7 +1860,7 @@ Full research background, citations, and methodology: [docs/research/](docs/rese
 
 ### For Engineering Teams
 
-**Standardize how your team works with AI across 12 provider integrations.** Whether your team uses Claude Code, Codex, Copilot, Cursor, Pi, or Warp, everyone can receive the same source workflows adapted to the provider's supported surfaces. The 35 enforcement rules prevent common AI mistakes: deleting tests to make them pass, fabricating citations, hard-coding tokens, or silently dropping features. Human-in-the-loop gates at phase transitions ensure no AI output reaches production without human review.
+**Standardize how your team works with AI across 14 provider integrations.** Whether your team uses Antigravity, Claude Code, Codex, Copilot, Cursor, Pi, or Warp, everyone can receive the same source workflows adapted to the provider's supported surfaces. The 35 enforcement rules prevent common AI mistakes: deleting tests to make them pass, fabricating citations, hard-coding tokens, or silently dropping features. Human-in-the-loop gates at phase transitions ensure no AI output reaches production without human review.
 
 ### For Platform Engineers
 

package/agentic/code/addons/aiwg-utils/docs/overview.md

@@ -8,7 +8,7 @@ Beyond the rules, aiwg-utils also provides:
 
 - **5 agents**: `context-regenerator`, `aiwg-developer`, `consortium-coordinator`, `self-debug`, `aiwg-steward`
 - **9 skills**: `project-awareness`, `schedule`, `soul-create`, `soul-validate`, `soul-enhance`, `soul-apply`, `soul-blend`, `aiwg-sync`, `aiwg-guide`
-- **Provider regeneration commands**: provider-native `aiwg-regenerate` prompts or skills across all 12 named integrations, adapted to each provider's command surface
+- **Provider regeneration commands**: provider-native `aiwg-regenerate` prompts or skills across all 13 named integrations, adapted to each provider's command surface
 - **Workspace commands**: `workspace-realign`, `workspace-prune-working`, `workspace-reset`
 - **@-mention tooling**: `mention-wire`, `mention-validate`, `mention-report`, `mention-lint`
 - **Hook management**: `hook-enable`, `hook-disable`, `hook-status`, `hook-regenerate`
@@ -59,7 +59,7 @@ Apply when: Telegram/group-chat bots, multi-bot rooms, direct-message business b
 
 #### native-ux-tools
 
-Use provider-native interaction tools (e.g., Claude Code's AskUserQuestion) for interactive questions rather than plain text. One question per interaction turn. The capability matrix covers all 12 named provider integrations and falls back to formatted markdown if native tools are unavailable.
+Use provider-native interaction tools (e.g., Claude Code's AskUserQuestion) for interactive questions rather than plain text. One question per interaction turn. The capability matrix covers all 13 named provider integrations and falls back to formatted markdown if native tools are unavailable.
 
 Apply when: interactive commands (`--interactive` flag), decision gates, intake wizards.
 

package/agentic/code/addons/aiwg-utils/rules/debug-source-not-output.md

@@ -180,7 +180,7 @@ When a stack trace or error points into build output, the required research acti
 ## Platform Applicability
 
 Applies universally across all AI coding platforms:
-- Claude Code, Codex, Copilot, Cursor, Warp, Factory, OpenCode, Windsurf
+- Every supported provider
 - Any agent performing debugging, code review, error tracing, or stack trace analysis
 
 ## Checklist

package/agentic/code/addons/aiwg-utils/rules/human-authorization.md

@@ -171,7 +171,7 @@ Both are required. A well-researched unauthorized action is still a violation.
 ## Platform Applicability
 
 This rule applies universally across all AI coding platforms:
-- Claude Code, Codex, Copilot, Cursor, Warp, Factory, OpenCode, Windsurf
+- Every supported provider
 - Any agent that modifies files, manages artifacts, or takes actions with lasting effects
 
 Authorization is not a platform feature. It is an agent responsibility.

package/agentic/code/addons/aiwg-utils/rules/instruction-comprehension.md

@@ -204,7 +204,7 @@ In the Thought-Action-Observation loop:
 ## Platform Applicability
 
 This rule applies universally across all AI coding platforms:
-- Claude Code, Codex, Copilot, Cursor, Warp, Factory, OpenCode, Windsurf
+- Every supported provider
 - Any agent that receives natural language instructions
 
 Following instructions is not a feature. It is the baseline expectation.

package/agentic/code/addons/aiwg-utils/rules/research-before-decision.md

@@ -192,7 +192,7 @@ A TAO iteration that goes Thought→Edit→Error without any Read/Search/Grep ac
 ## Platform Applicability
 
 This rule applies universally across all AI coding platforms:
-- Claude Code, Codex, Copilot, Cursor, Warp, Factory, OpenCode, Windsurf
+- Every supported provider
 - Any agent that modifies code or makes technical decisions
 
 Research is not optional. It is the minimum bar for competent assistance.

package/agentic/code/addons/aiwg-utils/rules/subagent-scoping.md

@@ -104,7 +104,7 @@ Consider subagents when non-trivial work decomposes into independent, bounded un
 
 ## Platform Applicability
 
-Universal across all AI coding platforms (Claude Code, Codex, Copilot, Cursor, Warp, Factory, OpenCode, Windsurf) and any agent that delegates work.
+Universal across every supported provider and any agent that delegates work.
 
 ## Checklist
 

package/agentic/code/addons/aiwg-utils/skills/aiwg-guide/SKILL.md

@@ -2,7 +2,7 @@
 namespace: aiwg
 name: aiwg-guide
 description: Contextual AIWG help — explains current version features, answers how-to questions, routes live queries to the steward
-platforms: [claude-code, codex, opencode, warp, cursor, windsurf, copilot, factory, openclaw, hermes]
+platforms: [all]
 
 ---
 

package/agentic/code/addons/aiwg-utils/skills/aiwg-issue/SKILL.md

@@ -102,7 +102,9 @@ node --version         # → Node version
 aiwg doctor            # workspace health snapshot (helpful)
 ```
 
-And — also required — the **provider** (the AIWG harness in use): one of `claude-code`, `hermes`, `codex`, `copilot`, `cursor`, `warp`, `factory`, `opencode`, `windsurf`, `openclaw`.
+And — also required — the **provider** (the AIWG harness in use): one of
+`claude-code`, `codex`, `copilot`, `cursor`, `factory`, `hermes`, `opencode`,
+`openclaw`, `openhuman`, `omp`, `pi`, `warp`, or `windsurf`.
 
 When you (the agent) draft a bug report on behalf of a user, **ask explicitly for each of the three required fields** if they aren't already in the conversation. Don't infer. The May-2026 jmagly→roctinam import sweep needed a correction round because the original GitHub report cited "Claude Code 2.1.137" but the actual harness in use was hermes — same bug, different behavior, wasted half a round-trip to clarify. Don't repeat that.
 

package/agentic/code/addons/aiwg-utils/skills/doc-consolidate/SKILL.md

@@ -7,7 +7,7 @@ commandHint:
   allowedTools: Read, Write, Bash, Glob, Grep
   category: utilities
   orchestration: true
-platforms: [claude-code]
+platforms: [all]
 
 ---
 

package/agentic/code/addons/rlm/docs/deployment-guide.md

@@ -6,7 +6,7 @@
 
 ## Overview
 
-The RLM (Recursive Language Models) addon provides recursive context decomposition for processing arbitrarily large codebases and document corpora through programmatic sub-agent delegation. This guide documents resource deployment across AIWG's 12 named provider integrations; runtime delegation depends on provider capability.
+The RLM (Recursive Language Models) addon provides recursive context decomposition for processing arbitrarily large codebases and document corpora through programmatic sub-agent delegation. This guide documents resource deployment across AIWG's 13 named provider integrations; runtime delegation depends on provider capability.
 
 ### What Gets Deployed
 
@@ -629,7 +629,7 @@ After deploying RLM to any provider:
 
 **Check**:
 1. Provider name spelled correctly (lowercase)
-2. The provider is one of AIWG's 12 named integrations (`aiwg help`)
+2. The provider is one of AIWG's 13 named integrations (`aiwg help`)
 3. Permissions to write to target directories
 4. For Codex: Home prompt directory and project `.agents/skills/` are writable
 

package/agentic/code/addons/rlm/docs/multi-provider-guide.md

@@ -6,7 +6,7 @@
 
 ## Overview
 
-The RLM (Recursive Language Models) addon supports running root agents and sub-agents across multiple AI providers. This enables cost optimization (e.g., Claude Opus root with Codex Mini sub-calls), provider-specific strengths (e.g., Claude for reasoning, OpenAI Codex for code generation), and resource deployment across AIWG's 12 named provider integrations. Runtime sub-agent execution still depends on the selected provider's capabilities.
+The RLM (Recursive Language Models) addon supports running root agents and sub-agents across multiple AI providers. This enables cost optimization (e.g., Claude Opus root with Codex Mini sub-calls), provider-specific strengths (e.g., Claude for reasoning, OpenAI Codex for code generation), and resource deployment across AIWG's 13 named provider integrations. Runtime sub-agent execution still depends on the selected provider's capabilities.
 
 ## Model Mapping for RLM
 

package/agentic/code/addons/semantic-memory/skills/memory-ingest/SKILL.md

@@ -3,7 +3,7 @@ name: memory-ingest
 description: Ingest a source into any consumer's semantic memory by reading the topology contract
 namespace: aiwg
 category: kernel
-platforms: [claude, copilot, cursor, factory, windsurf, warp, codex, opencode, openclaw, hermes]
+platforms: [all]
 triggers:
   - "help my AI remember what we decided"
   - "help me preserve project decisions between sessions"

package/agentic/code/addons/semantic-memory/skills/memory-lint/SKILL.md

@@ -3,7 +3,7 @@ name: memory-lint
 description: Health-check any consumer's semantic memory by composing structural and domain-specific checks
 namespace: aiwg
 category: kernel
-platforms: [claude, copilot, cursor, factory, windsurf, warp, codex, opencode, openclaw, hermes]
+platforms: [all]
 ---
 
 # memory-lint

package/agentic/code/addons/semantic-memory/skills/memory-log-append/SKILL.md

@@ -3,7 +3,7 @@ name: memory-log-append
 description: Append a structured event to a consumer's semantic memory log
 namespace: aiwg
 category: kernel
-platforms: [claude, copilot, cursor, factory, windsurf, warp, codex, opencode, openclaw, hermes]
+platforms: [all]
 ---
 
 # memory-log-append

package/agentic/code/addons/semantic-memory/skills/memory-log-render/SKILL.md

@@ -3,7 +3,7 @@ name: memory-log-render
 description: Generate a human-readable Markdown view from a consumer's JSON Lines event log
 namespace: aiwg
 category: kernel
-platforms: [claude, copilot, cursor, factory, windsurf, warp, codex, opencode, openclaw, hermes]
+platforms: [all]
 ---
 
 # memory-log-render

package/agentic/code/addons/semantic-memory/skills/memory-query-capture/SKILL.md

@@ -3,7 +3,7 @@ name: memory-query-capture
 description: Capture query synthesis as durable pages in semantic memory
 namespace: aiwg
 category: kernel
-platforms: [claude, copilot, cursor, factory, windsurf, warp, codex, opencode, openclaw, hermes]
+platforms: [all]
 ---
 
 # memory-query-capture

package/agentic/code/frameworks/sdlc-complete/skills/doc-consolidate/SKILL.md

@@ -7,7 +7,7 @@ commandHint:
   allowedTools: Read, Write, Bash, Glob, Grep
   category: utilities
   orchestration: true
-platforms: [claude-code]
+platforms: [all]
 
 ---
 

package/agentic/code/frameworks/sdlc-complete/templates/antigravity/AGENTS.md.aiwg-template

@@ -0,0 +1,8 @@
+# Google Antigravity CLI workspace context
+
+<!-- AIWG SDLC Framework Integration -->
+
+Read and follow [WORKSPACE.md](./WORKSPACE.md) first.
+Then read [AIWG.md](./AIWG.md) for AIWG discovery, quickrefs, and framework routing.
+
+These are explicit reading instructions. Plain Markdown links are not claimed to auto-load.

package/agentic/code/plugins/agent-loop/.claude-plugin/plugin.json

@@ -1,6 +1,6 @@
 {
   "name": "agent-loop",
-  "version": "2026.9.2",
+  "version": "2026.9.4",
   "description": "Iterative AI coding loops with automatic recovery - iteration beats perfection",
   "author": {
     "name": "AIWG Contributors",

package/agentic/code/plugins/agent-persistence/.claude-plugin/plugin.json

@@ -1,6 +1,6 @@
 {
   "name": "agent-persistence",
-  "version": "2026.9.2",
+  "version": "2026.9.4",
   "description": "Human-in-the-loop (HITL) gate definitions for agent operations. Provides reusable YAML gates for destructive-action authorization, false-positive override, and recovery escalation. Loaded by agent loops and orchestrators that need bounded human checkpoints.",
   "author": {
     "name": "AIWG Contributors",

package/agentic/code/plugins/agentic-installer/.claude-plugin/plugin.json

@@ -1,6 +1,6 @@
 {
   "name": "agentic-installer",
-  "version": "2026.9.2",
+  "version": "2026.9.4",
   "description": "Structured YAML manifests for reproducible, agent-driven software installation across platforms. Scripts are the primary artifact; agentic instructions handle adaptation and recovery.",
   "author": {
     "name": "AIWG Contributors",

package/agentic/code/plugins/aiwg-dev/.claude-plugin/plugin.json

@@ -1,6 +1,6 @@
 {
   "name": "aiwg-dev",
-  "version": "2026.9.2",
+  "version": "2026.9.4",
   "description": "Developer tooling for building AIWG itself and AIWG addons/frameworks. Not for end users — install explicitly with `aiwg use aiwg-dev`.",
   "author": {
     "name": "AIWG Contributors",

package/agentic/code/plugins/aiwg-evals/.claude-plugin/plugin.json

@@ -1,6 +1,6 @@
 {
   "name": "aiwg-evals",
-  "version": "2026.9.2",
+  "version": "2026.9.4",
   "description": "Automated evaluation framework for agent quality, based on KAMI benchmark methodology and ReAct patterns",
   "author": {
     "name": "AIWG Contributors",

package/agentic/code/plugins/aiwg-evals/README.md

@@ -15,6 +15,14 @@ AIWG Evals provides:
 2. **Workflow Scenarios**: Test multi-agent orchestration
 3. **Archetype Detection**: Identify failure pattern susceptibility
 4. **Quality Reports**: Track agent health over time
+5. **Integrity-Gated Evidence**: Distinguish smoke diagnostics from verified
+   evals with uncertainty, paired baselines, compromise labels, and
+   PROMOTE/HOLD/ROLLBACK decisions
+
+> A dimension score is a smoke diagnostic unless the report includes
+> `sample_n`, uncertainty, `integrity_state`, and `trusted_score_source`.
+> Detected `test_edit`, `scorer_edit`, `fixture_edit`, `metric_leakage`, or
+> `unknown` compromise labels prevent promotion regardless of score.
 
 ## Installation
 

package/agentic/code/plugins/auto-memory/.claude-plugin/plugin.json

@@ -1,6 +1,6 @@
 {
   "name": "auto-memory",
-  "version": "2026.9.2",
+  "version": "2026.9.4",
   "description": "Automatic memory seed templates for AIWG projects",
   "author": {
     "name": "AIWG Contributors",

package/agentic/code/plugins/browser-control/.claude-plugin/plugin.json

@@ -1,6 +1,6 @@
 {
   "name": "browser-control",
-  "version": "2026.9.2",
+  "version": "2026.9.4",
   "description": "Drive a real, user-installed Chromium-derived browser from an AIWG agent. Wraps @playwright/mcp with a setup wizard, health-check doctor, allow-list enforcement, and an opinionated agent persona that respects authorization and audit boundaries.",
   "author": {
     "name": "AIWG Contributors",

package/agentic/code/plugins/codex-sdlc/skills/doc-consolidate/SKILL.md

@@ -7,7 +7,7 @@ commandHint:
   allowedTools: Read, Write, Bash, Glob, Grep
   category: utilities
   orchestration: true
-platforms: [claude-code]
+platforms: [all]
 
 ---
 

package/agentic/code/plugins/color-palette/.claude-plugin/plugin.json

@@ -1,6 +1,6 @@
 {
   "name": "color-palette",
-  "version": "2026.9.2",
+  "version": "2026.9.4",
   "description": "Color palette selection and review tooling with color theory fundamentals, current trend research, and accessibility analysis.",
   "author": {
     "name": "AIWG Contributors",

package/agentic/code/plugins/compound-memory/.claude-plugin/plugin.json

@@ -1,6 +1,6 @@
 {
   "name": "compound-memory",
-  "version": "2026.9.2",
+  "version": "2026.9.4",
   "description": "Governed orchestration across AIWG line memory, semantic memory, and llm-wiki.",
   "author": {
     "name": "AIWG Contributors",

package/agentic/code/plugins/compound-memory/docs/operator-guide.md

@@ -1,5 +1,10 @@
 # Compound Memory operator guide
 
+Compound Memory remains independently supported. Dataset Intelligence may
+register source/run evidence, but does not replace promotion, review, bounded
+context, or maintenance semantics. See the [dataset migration
+guide](../dataset-intelligence/migration-guide.md).
+
 ## Activate
 
 ```bash

package/agentic/code/plugins/context-curator/.claude-plugin/plugin.json

@@ -1,6 +1,6 @@
 {
   "name": "context-curator",
-  "version": "2026.9.2",
+  "version": "2026.9.4",
   "description": "Context curation and distractor filtering for production-grade agent reliability",
   "author": {
     "name": "AIWG Contributors",

package/agentic/code/plugins/daemon/.claude-plugin/plugin.json

@@ -1,6 +1,6 @@
 {
   "name": "daemon",
-  "version": "2026.9.2",
+  "version": "2026.9.4",
   "description": "Persistent daemon session capabilities: Concierge front-end, session memory, scheduler integration",
   "author": {
     "name": "AIWG Contributors",

package/agentic/code/plugins/daemon/agents/concierge.md

@@ -112,8 +112,8 @@ Never expose file paths in the confirmation response.
 
 ## References
 
-- @$AIWG_ROOT/agentic/code/addons/daemon/behaviors/concierge.behavior.md — Behavior definition (full teach mode spec)
-- @$AIWG_ROOT/agentic/code/addons/daemon/rules/daemon-interaction.md — Interaction rules
+- @$AIWG_ROOT/${CLAUDE_PLUGIN_ROOT}/behaviors/concierge.behavior.md — Behavior definition (full teach mode spec)
+- @$AIWG_ROOT/${CLAUDE_PLUGIN_ROOT}/rules/daemon-interaction.md — Interaction rules
 - @$AIWG_ROOT/docs/daemon-guide.md — Daemon architecture
 - @$AIWG_ROOT/agentic/code/addons/prose-integration/skills/prose-run/SKILL.md — OpenProse runner for teach mode delegation
 - @$AIWG_ROOT/agentic/code/addons/prose-integration/skills/prose-detect/SKILL.md — OpenProse detection

package/agentic/code/plugins/doc-intelligence/.claude-plugin/plugin.json

@@ -1,6 +1,6 @@
 {
   "name": "doc-intelligence",
-  "version": "2026.9.2",
+  "version": "2026.9.4",
   "description": "Documentation intelligence addon for scraping, extraction, analysis, and generated-docs auditing.",
   "author": {
     "name": "AIWG Contributors",

package/agentic/code/plugins/droid-bridge/.claude-plugin/plugin.json

@@ -1,6 +1,6 @@
 {
   "name": "droid-bridge",
-  "version": "2026.9.2",
+  "version": "2026.9.4",
   "description": "MCP server bridging Claude Code to Factory Droid for batch operations and automated fixes",
   "author": {
     "name": "AIWG Contributors",

package/agentic/code/plugins/forensics/.claude-plugin/plugin.json

@@ -1,6 +1,6 @@
 {
   "name": "forensics",
-  "version": "2026.9.2",
+  "version": "2026.9.4",
   "description": "Digital forensics and incident response framework with 14 specialized agents covering target profiling, evidence acquisition, log/memory/container/cloud analysis, IOC extraction, and reporting.",
   "author": {
     "name": "AIWG Contributors",

package/agentic/code/plugins/guided-implementation/.claude-plugin/plugin.json

@@ -1,6 +1,6 @@
 {
   "name": "guided-implementation",
-  "version": "2026.9.2",
+  "version": "2026.9.4",
   "description": "Bounded iteration control for autonomous issue-to-code implementation",
   "author": {
     "name": "AIWG Contributors",

package/agentic/code/plugins/guided-implementation/docs/overview.md

@@ -47,7 +47,7 @@ Task: Implement JWT token validation
 
 Attempts:
 - Iter 1: Test failed — undefined token variable
-- Iter 2: Test failed — token missing userId field
+- Iter 2: Test failed — token missing userId field  
 - Iter 3: Test failed — userId format mismatch (string vs number)
 
 Pattern detected: userId type inconsistency between implementation and test

package/agentic/code/plugins/hooks/.claude-plugin/plugin.json

@@ -1,6 +1,6 @@
 {
   "name": "hooks",
-  "version": "2026.9.2",
+  "version": "2026.9.4",
   "description": "Claude Code hooks for workflow tracing and observability.",
   "author": {
     "name": "AIWG Contributors",

package/agentic/code/plugins/knowledge-base/.claude-plugin/plugin.json

@@ -1,6 +1,6 @@
 {
   "name": "knowledge-base",
-  "version": "2026.9.2",
+  "version": "2026.9.4",
   "description": "Knowledge-base / wiki framework with skills for ingestion, query, and lint. Build a structured, citable corpus with cross-references and concept pages.",
   "author": {
     "name": "AIWG Contributors",

package/agentic/code/plugins/line-memory/.claude-plugin/plugin.json

@@ -1,6 +1,6 @@
 {
   "name": "line-memory",
-  "version": "2026.9.2",
+  "version": "2026.9.4",
   "description": "Bounded, plain-text project memory with least-recently-used retention.",
   "author": {
     "name": "AIWG Contributors",

package/agentic/code/plugins/llm-wiki/.claude-plugin/plugin.json

@@ -1,6 +1,6 @@
 {
   "name": "llm-wiki",
-  "version": "2026.9.2",
+  "version": "2026.9.4",
   "description": "LLM-maintained markdown wiki that compounds as sources are ingested. Thin topology declaration on the semantic memory kernel — ships schema profiles and page templates, no custom mechanics.",
   "author": {
     "name": "AIWG Contributors",

package/agentic/code/plugins/marketing/.claude-plugin/plugin.json

@@ -1,6 +1,6 @@
 {
   "name": "marketing",
-  "version": "2026.9.2",
+  "version": "2026.9.4",
   "description": "Marketing automation framework with 37 specialized agents for campaign management.",
   "author": {
     "name": "AIWG Contributors",

package/agentic/code/plugins/media-curator/.claude-plugin/plugin.json

@@ -1,6 +1,6 @@
 {
   "name": "media-curator",
-  "version": "2026.9.2",
+  "version": "2026.9.4",
   "description": "Media archive management framework with 7 specialized agents for source discovery, acquisition (yt-dlp / Internet Archive / Bandcamp), quality assessment, metadata tagging, and provenance tracking.",
   "author": {
     "name": "AIWG Contributors",

package/agentic/code/plugins/nlp-prod/.claude-plugin/plugin.json

@@ -1,6 +1,6 @@
 {
   "name": "nlp-prod",
-  "version": "2026.9.2",
+  "version": "2026.9.4",
   "description": "Tools, skills, and CLI extensions for designing and productionizing LLM inference pipelines. Pattern-guided, slim by default, eval-first.",
   "author": {
     "name": "AIWG Contributors",

package/agentic/code/plugins/ops/.claude-plugin/plugin.json

@@ -1,6 +1,6 @@
 {
   "name": "ops",
-  "version": "2026.9.2",
+  "version": "2026.9.4",
   "description": "Operational infrastructure framework with 12 specialized agents covering incident response, runbook execution, fleet inventory, certificate lifecycle, and disaster recovery planning.",
   "author": {
     "name": "AIWG Contributors",

package/agentic/code/plugins/ops/agents/ops-runbook-executor.md

@@ -18,7 +18,7 @@ Execute operational runbooks with structured verification, safety enforcement, a
 - Verify expected output at each step before proceeding
 - Gate destructive operations (require explicit human approval)
 - Detect interactive commands and flag for human execution
-- Generate audit trail of all commands run, outputs received, and files modified
+- Generate governed minimum-sufficient evidence for commands, outcomes, bounded output excerpts, and file changes
 - Rollback on failure if rollback procedure is documented
 - Post progress updates to issue thread if issue context is provided
 
@@ -28,11 +28,13 @@ Execute operational runbooks with structured verification, safety enforcement, a
 - ALWAYS assess blast radius before destructive operations
 - IF a step fails, check troubleshooting section before retrying
 - LIMIT retries to 3 per step — escalate after that
-- RECORD every command executed and its output in the audit trail
+- RECORD each command/result with bounded redacted excerpts and digests; full raw capture requires an explicit reason and short-lived raw tier
+- ALWAYS invoke the ops evidence boundary before an audit/artifact write, agent response containing collected output, or issue/PR/comment submission
+- NEVER submit a denied payload; use the separately gated sanitized summary when the boundary returns one
 - NEVER apply one host's runbook to a different host without explicit confirmation
 
 ## Output Format
-After execution, produce an audit trail:
+After execution, prepare the audit trail through `aiwg ops evidence prepare` (or the `aiwg/governance` API), then produce:
 | Step | Command | Expected | Actual | Status |
 |------|---------|----------|--------|--------|
 | 1 | {cmd} | {expected} | {actual} | PASS/FAIL |