Add Anchor plugin (Community / Development & Workflow) (#144)

* Add Anchor plugin (Community / Development & Workflow)

Anchor is a cross-CLI (Claude Code + Codex CLI) engineering-discipline pack
that bundles a skill, 11 slash commands, and 4 safety hooks. Highlights:

- task-scope anchoring on the user's exact phrasing
- anti-drift braking when the model tries to go out of scope
- optional autonomous mode (Stop hook keeps the model going while tasks remain)
- end-to-end verification gate (refuses to declare done before a real E2E)
- multi-pass vulnerability scanning + condition-based codex review on big diffs
- pitfall writeback to the project's CLAUDE.md
- PreToolUse hook blocking irreversible bash patterns (hard-reset, force-push,
  schema-level drops, mkfs, dd-to-device, recursive-777, curl-pipe-bash) with
  a shell-aware segment + safe-list parser so echo/grep literals don't false-trip

Bundle:
- plugins/biefan/anchor/.codex-plugin/plugin.json  (name, version 1.3.1,
  license MIT, interface.composerIcon, keywords)
- plugins/biefan/anchor/assets/icon.svg            (711 bytes, 512x512)
- plugins/biefan/anchor/LICENSE                    (MIT)
- plugins/biefan/anchor/README.md                  (points at the full repo)

README entry inserted alphabetically between AgentOps and Antigravity.

Upstream:  https://github.com/biefan/anchor
Release:   https://github.com/biefan/anchor/releases/tag/v1.3.1
License:   MIT
CI:        green (shellcheck + jsonlint + install.sh smoke test)

* anchor: bump bundle plugin.json to v1.3.2 (matches upstream release)

Author: biefan

README.md

@@ -123,6 +123,7 @@ Third-party plugins built by the community. [PRs welcome](#contributing)!
 - [Aegis](https://github.com/GanyuanRan/Aegis) - An agentic skills framework & software development methodology that works: planning, TDD, debugging, and collaboration workflows.
 - [Agentizer](https://github.com/Humiris/wwa-transform) - Turn any website into an AI-powered agentfront with split-pane
 - [AgentOps](https://github.com/boshu2/agentops) - DevOps layer for coding agents with flow, feedback, and memory that compounds between sessions.
+- [Anchor](https://github.com/biefan/anchor) - Engineering discipline pack for Claude Code & Codex CLI with task-scope locking, anti-drift braking, condition-based codex review, project-CLAUDE.md pitfall writeback, and PreToolUse hooks that block irreversible bash patterns.
 - [Antigravity Workspace Template](https://github.com/study8677/antigravity-workspace-template) - Multi-agent codebase knowledge graph generator with context-aware planning and automatic scope management — turns codebases into coherent agent workspaces.
 - [Archcore](https://github.com/archcore-ai/plugin) - Gives coding agents the architecture, rules, and prior decisions of the repo via skills, hooks, and MCP — so new changes land where the project says they belong across Claude Code, Cursor, and Codex CLI.
 - [Bring Your AI Migration Auditor](https://github.com/unitedideas/bringyour-mcp) - Read-only Codex plugin for auditing Claude Code to Codex migrations before Codex edits code. Checks AGENTS.md/CLAUDE.md scope, hooks, MCP config, skills, secret references, and validation notes.

plugins/biefan/anchor/.codex-plugin/plugin.json

@@ -0,0 +1,29 @@
+{
+  "name": "anchor",
+  "version": "1.3.2",
+  "description": "Engineering discipline pack for Claude Code & Codex CLI: a skill + 11 slash commands + 4 safety hooks that keep AI on-task on long tasks, enforce E2E verification, multi-pass vulnerability scanning, condition-based codex review, and project-CLAUDE.md pitfall writeback. Cross-CLI by design.",
+  "author": {
+    "name": "biefan",
+    "url": "https://github.com/biefan"
+  },
+  "homepage": "https://github.com/biefan/anchor",
+  "repository": "https://github.com/biefan/anchor",
+  "license": "MIT",
+  "keywords": [
+    "engineering-discipline",
+    "skill",
+    "hooks",
+    "slash-commands",
+    "claude-code",
+    "codex-cli",
+    "anti-drift",
+    "autonomous-mode",
+    "e2e-verification",
+    "pitfall-writeback"
+  ],
+  "interface": {
+    "displayName": "Anchor",
+    "shortDescription": "Engineering discipline for AI coding agents — anchor scope, run E2E, write pitfalls back.",
+    "composerIcon": "./assets/icon.svg"
+  }
+}

plugins/biefan/anchor/LICENSE

@@ -0,0 +1,21 @@
+MIT License
+
+Copyright (c) 2026 biefan
+
+Permission is hereby granted, free of charge, to any person obtaining a copy
+of this software and associated documentation files (the "Software"), to deal
+in the Software without restriction, including without limitation the rights
+to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
+copies of the Software, and to permit persons to whom the Software is
+furnished to do so, subject to the following conditions:
+
+The above copyright notice and this permission notice shall be included in all
+copies or substantial portions of the Software.
+
+THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
+AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
+OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+SOFTWARE.

plugins/biefan/anchor/README.md

@@ -0,0 +1,37 @@
+# Anchor
+
+Engineering discipline pack for Claude Code & Codex CLI: a skill + 11 slash commands + 4 safety hooks that keep AI on-task on long tasks, enforce E2E verification, multi-pass vulnerability scanning, condition-based codex review, and project-CLAUDE.md pitfall writeback.
+
+- **Upstream repo**: https://github.com/biefan/anchor
+- **Release tag**: v1.3.1
+- **License**: MIT
+
+## What's in the bundle
+
+This is the mirror bundle for awesome-codex-plugins. For the full repo (including hook scripts, references, evals, and install.sh), see the upstream link above. To install:
+
+```bash
+git clone https://github.com/biefan/anchor.git ~/anchor
+cd ~/anchor
+./install.sh
+```
+
+`install.sh` is cross-CLI: it copies the skill + 7 commands to `~/.claude/` and (if codex CLI is on PATH) to `~/.codex/`, and auto-merges hooks into `~/.claude/settings.json` (idempotent, with backup, `--no-hooks` to skip).
+
+## What it does
+
+| Layer | Mechanism |
+|---|---|
+| **Skill (soft rules)** | `SKILL.md` — 7 core rules: clarify intent, lock task scope, read project contracts, smallest-correct-diff, parallelize agents, condition-based codex review, pitfall writeback. |
+| **11 slash commands** | `/ec /lock /pit /scan /done /next /recap /init-claude-md /status /ship /diff /cleanup` |
+| **4 hooks (hard enforcement)** | `SessionStart` injects project state; `Stop` blocks "stop" in autonomous mode while tasks remain; `PreToolUse` blocks irreversible bash patterns (force-push, hard-reset, drop, mkfs, dd-to-device, recursive-777, curl-pipe-bash); `PostToolUse` runs the language-appropriate linter after Edit/Write. |
+
+## Validation
+
+- CI: green (`shellcheck` + `jsonlint` + `install.sh` smoke test on clean Ubuntu)
+- Stress tests: `evals/stress/03-debug-failing-tests` scored 6/1/1 (pass/fail/N/A) via codex-as-judge
+- See [CHANGELOG.md](https://github.com/biefan/anchor/blob/main/CHANGELOG.md) for release notes across v1.0 - v1.3.1
+
+## License
+
+MIT