Agent Plugins Marketplace
All plugins

anti-hall

v0.104.0

Verify-first discipline for Codex. Bundles Codex-native skills (including session handover — a comprehensive handoff so a fresh session can resume without re-deriving or guessing anything) plus an installer for the Codex hook subset that can be enforced today: session/prompt nudges, a codex-availability PATH probe, shell-command guards, Stop guards, task discipline, and speculation checks. Claude Code plugin files are kept separate and unchanged.

CodexClaude Code13 Skills

By Mohammed TalasLicense: MIT1 GitHub starsUpdated 1 hour ago

Directory evidence

Runtimes
Codex and Claude Code
Parsed components
13 skill or MCP entries
Source updated
Sep 24, 2026
Manifest status
Canonical path parsed

The directory validates manifest shape and source location. It does not execute the plugin or provide a security endorsement. Review the indexing methodology

Install anti-hall for Codex and Claude Code

Installs for the current user
codex plugin marketplace add talas9/anti-hall
codex plugin marketplace upgrade anti-hall
codex plugin add anti-hall@anti-hall

Paste and run these commands in a terminal with Codex. They add and refresh the anti-hall catalog, then install this plugin.

Compatibility: the page URL and API slug “anti-hall” remain stable.

  • Codex: anti-hall@agent-plugin-marketplaceanti-hall@anti-hall

The installer fetches third-party code from the source repository shown on this page. This directory validates manifest structure and source location, but does not perform a security audit; review the manifest, components, and source before installing.

Get the source manually
git clone https://github.com/talas9/anti-hall

Clone the source repository, then follow its setup instructions to add the plugin to a compatible client. The plugin root is plugins/anti-hall/.

Plugin files

plugins/anti-hall/
├── .codex-plugin/plugin.json
├── .claude-plugin/plugin.json
├── skills/activate/SKILL.md
├── skills/deadly-loop/SKILL.md
├── skills/deadly-loop-multi/SKILL.md
├── skills/defects/SKILL.md
├── skills/devswarm/SKILL.md
├── skills/doctor/SKILL.md
├── skills/flutter-debug/SKILL.md
├── skills/handover/SKILL.md
├── skills/install-statusline/SKILL.md
├── skills/root-cause/SKILL.md
├── skills/ship-it/SKILL.md
├── skills/system-briefing/SKILL.md
└── skills/update/SKILL.md

Included Skills13

activateskills/activate/SKILL.md

One-shot idempotent anti-hall setup. Checks statusline installation, reports model-routing state, writes a sentinel so it doesn't repeat. Use when the user says "activate anti-hall", "set up anti-hall", "run first-time setup", or "anti-hall activate". NOT auto-run — user-invoked only.

deadly-loopskills/deadly-loop/SKILL.md

Iterative TRIO-debate + fix-wave protocol for hardening non-trivial PRs and changes before merge. Spawns a parallel Reviewer + Auditor + Critic trio, dispatches fix waves, loops until convergence (zero NEW P0 or P1 blockers). Use before merging anything with cross-file risk, cross-PR coordination, security-sensitive changes (auth, signing, redaction, prompt-injection defense), schema migrations, production-data touches, shell scripts, CI/workflow YAML, cross-repo coordination, or LLM prompt work — anywhere a silent failure would be expensive. Born from a real project where 7 iteration rounds caught 30+ bugs that solo review would have missed.

deadly-loop-multiskills/deadly-loop-multi/SKILL.md

Run a DOUBLE / TRIPLE / QUADRUPLE deadly loop — the deadly-loop TRIO (Sonnet Reviewer + Opus Auditor + Codex Critic) multiplied N× in parallel, auditing a target (whole repo, a diff, or named files) with diversified lenses, then dedup + synthesize, optionally fix-wave and re-converge. Use when the user says "double deadly loop", "triple deadly loop", "quadruple deadly loop", "deep multi-agent review", "have multiple flagship + codex agents review everything", or wants a heavier review than the standard 1× deadly-loop. Multiplier — double=2× the trio (6 agents), triple=3× (9), quadruple=4× (12).

defectsskills/defects/SKILL.md

File, list, show, and rule on anti-hall's own defect reports via the durable, home-scoped, two-way defect channel. Use when the user says "file an anti-hall bug", "did they fix my report", "anti-hall defects", "report this to the maintainer", or "check my defect reports".

devswarmskills/devswarm/SKILL.md

Explain and activate anti-hall's optional DevSwarm integration — the hivecontrol reference KB, the workspace-tier orchestration doctrine (a Primary's top fan-out tier is a child workspace, not a subagent — doctrine + guard redirects shipped, no mechanical classifier), and the shipped layered recovery model (child self-report → supervisor poke → escalate-to-parent, automatic path NEVER kills) plus the on-demand devswarm-recover CLI (the only path that ever kills). Use when the user asks "explain the anti-hall DevSwarm integration", "how do I activate the DevSwarm supervisor", "what DevSwarm addons does anti-hall have", "tune the liveness supervisor", "recover a stuck DevSwarm workspace", or anything about hivecontrol / DevSwarm workspaces from an anti-hall angle.

doctorskills/doctor/SKILL.md

Health-check AND repair anti-hall — confirm Node is found, every hook is present + syntax-valid, the guards actually fire (live behavioral self-tests on git-guard / command-guard / edit-guard / swarm-guard / model-routing-guard), the statusline is installed, and (in repair mode) auto-apply safe fixes. Use when the user asks "is anti-hall working / active / running", "check the hooks", "anti-hall doctor", "repair anti-hall", "fix the daemon", "are the guards on", or after install/update to verify everything is live.

flutter-debugskills/flutter-debug/SKILL.md

Drive a Flutter app in debug mode and close the fix loop — run with agent-controlled hot reload, drive the UI with semantic taps + agent-visible screenshots, watch runtime errors / logs / VM service, then reproduce → read error → root-cause → fix → hot reload → visually re-verify. Use when the user says "debug my Flutter app", "drive the simulator/emulator", "reproduce this bug in the app", "fix this and verify it in the UI", "hot reload and check", or wants an end-to-end Flutter debug loop. iOS fully; Android taps/screenshots VERIFIED on emulator (FP7 2026-06-11).

handoverskills/handover/SKILL.md

Use when the user says "prepare a handover", "handover", "write a handoff", "save session state", "hand this off", before /compact or /clear, at session end, or before ending work on a long task — writes a comprehensive, organized, minimal-but-lossless session handover so a fresh session can resume without re-deriving or guessing anything.

install-statuslineskills/install-statusline/SKILL.md

Install or enable the anti-hall statusline (rich line-1 statusline + phase bar) in the current repo or globally. Use when the user asks to "install the statusline", "add the bar", "enable the phase bar here", "set up the statusline in this project", or "show the statusline everywhere". Writes the statusLine entry into the correct settings file (user scope by default, project-local on request) and reminds the user to restart.

root-causeskills/root-cause/SKILL.md

Evidence-driven debugging discipline. Use when investigating ANY bug, failure, crash, alert, flaky test, or unexpected behavior, before proposing or applying a fix. Enforces no-cause-no-fix - collect evidence, trace the full sequence, instrument when evidence is missing, prove the original and root cause (not the surface symptom), then fix and verify. Trigger on "why is X failing", "debug Y", "this error", "it crashes", "investigate Z", or any symptom that does not yet have a proven mechanism.

ship-itskills/ship-it/SKILL.md

One lean, anti-hall-native workflow for shipping any change correctly — from a one-line fix to a multi-phase feature. Scales rigor to blast radius (S/M/L tiers). Brainstorm + plan happen IN PLAN MODE (ExitPlanMode is the approval gate); the plan is hardened with the deadly-loop BEFORE any code; on large work the disjoint build phases and the per-phase deadly-loop fan out as a Workflow swarm; each phase is verified with fresh evidence and a vacuous-test guard, then hardened until zero NEW P0s or P1s. Standalone — depends only on Claude Code built-ins (plan mode, the Workflow tool) plus anti-hall's own deadly-loop and always-on guards. Use when the user says "let's build X" / "implement Y" / "fix Z" and you want it done right the first time. Replaces feature-launch.

system-briefingskills/system-briefing/SKILL.md

Brief the agent that installs or operates anti-hall on how the whole system works — a DERIVED, live enumeration of every installed hook (grouped by event, each with its own one-line purpose), the shipped skills, the DevSwarm coordination substrate (mechanical triggers, store, CLI, auto-safe migration), and a docs/KB map. Generated from the actual hooks.json and the files on disk, never a hardcoded list, so it can't drift. Use when the user asks "brief me on anti-hall", "what's in this build", "how does the whole system work", "give me the anti-hall system overview", "what hooks/skills/substrate ship here", or when onboarding an agent to install or run anti-hall. For "is it actually working / do the guards fire", use the `doctor` skill instead.

updateskills/update/SKILL.md

Update anti-hall to the latest released version and show the changelog delta. Use when the user says "update anti-hall", "/anti-hall:update", "upgrade the plugin", "is anti-hall up to date", or "check for an anti-hall update". Fast-forward-pulls the marketplace clone, mirrors the new version into the plugin cache, prints what changed, then has the user reload in-session via /reload-plugins (rarely, a harness build may require a restart — the skill says so when relevant).

Plugin manifests2

plugins/anti-hall/.codex-plugin/plugin.json
{
  "name": "anti-hall",
  "version": "0.104.0",
  "description": "Verify-first discipline for Codex. Bundles Codex-native skills (including session handover — a comprehensive handoff so a fresh session can resume without re-deriving or guessing anything) plus an installer for the Codex hook subset that can be enforced today: session/prompt nudges, a codex-availability PATH probe, shell-command guards, Stop guards, task discipline, and speculation checks. Claude Code plugin files are kept separate and unchanged.",
  "author": {
    "name": "Mohammed Talas",
    "url": "https://github.com/talas9"
  },
  "homepage": "https://github.com/talas9/anti-hall",
  "repository": "https://github.com/talas9/anti-hall",
  "license": "MIT",
  "keywords": [
    "verification",
    "anti-hallucination",
    "codex",
    "quality",
    "guards",
    "workflow"
  ],
  "skills": "./codex/skills/",
  "interface": {
    "displayName": "Anti-Hall",
    "shortDescription": "Verify-first guardrails and workflows for Codex",
    "longDescription": "Anti-Hall brings verify-first discipline, command/git guards, task reminders, and Codex-native workflow guidance to Codex. Some Claude edit-time and subagent lifecycle hooks have no direct Codex equivalent; those are exposed as skill/workflow protocols instead of pretending to hard-block.",
    "developerName": "Mohammed Talas",
    "category": "Developer Tools",
    "capabilities": [
      "Read",
      "Write"
    ],
    "websiteURL": "https://github.com/talas9/anti-hall",
    "privacyPolicyURL": "https://github.com/talas9/anti-hall",
    "termsOfServiceURL": "https://github.com/talas9/anti-hall/blob/main/LICENSE",
    "defaultPrompt": [
      "Use anti-hall to verify this change before claiming it is done"
    ],
    "brandColor": "#111827",
    "screenshots": []
  },
  "hooks": "./codex/hooks/hooks.json"
}
plugins/anti-hall/.claude-plugin/plugin.json
{
  "$schema": "https://json.schemastore.org/claude-code-plugin-manifest.json",
  "name": "anti-hall",
  "displayName": "Anti-Hallucination Verify-First",
  "version": "0.104.0",
  "description": "Verify-first discipline for Claude Code. SessionStart injects the full Iron-Law + rationalization-table protocol, the always-on SCOPE & FIDELITY discipline (simplest sufficient solution, intent over letter, confirm before expanding scope, match rigor to blast radius, finish asked work / drop nothing), and a skill primer (and re-injects it after compaction, since SessionStart re-fires with source=compact); a per-turn UserPromptSubmit hook adds a short varying nudge. Skills: root-cause debugging, orchestration (non-blocking swarm), ship-it (one lean plan-mode + Workflow-swarm + deadly-loop-gated workflow scaled S/M/L to blast radius), deadly-loop, handover (comprehensive session handoff so a fresh session can resume without re-deriving or guessing anything). Ships api-guard (blocks fabricated stdlib/builtin APIs in code, verified against the installed python3/node; opt-in 3rd-party via ANTIHALL_API_GUARD_THIRDPARTY=1), git-guard (blocks force-push + AI self-credit in commit trailers AND gh pr/issue/release bodies), task discipline, and a monorepo/simple statusline (plus a repo-root AGENTS.md Codex mirror for clone-based use — not bundled by /plugin install). Also ships an OPT-IN mcp-reaper companion (macOS + Linux) — a pure-Node background job that kills ONLY orphaned MCP-server processes (parent already died); install via `node plugins/anti-hall/companion/install-reaper.js` (Windows is a documented no-op). All hooks and the statusline are pure Node (built-ins only) and run unchanged on macOS and Linux; Windows is untested and not officially supported. REQUIRES Node.js >= 22 on PATH: hooks are launched as `node <hook>.js`; without Node, the hooks (including the git-guard safety) silently do not run (intentional — a shell-based preflight cannot reach stock Windows systems, so we omit it by design). Verify with `node --version` before relying on protections. Includes a separate Codex-native port under plugins/anti-hall/codex with a .codex-plugin manifest, Codex hook-subset installer, OMX-oriented skills, context-conservation and ship-it protocols; Claude hook files remain unchanged.",
  "author": {
    "name": "Mohammed Talas",
    "url": "https://github.com/talas9"
  },
  "homepage": "https://github.com/talas9/anti-hall",
  "repository": "https://github.com/talas9/anti-hall",
  "license": "MIT",
  "keywords": [
    "verification",
    "anti-hallucination",
    "quality",
    "discipline",
    "debugging",
    "root-cause",
    "orchestration",
    "deadly-loop",
    "ship-it",
    "code-simplification",
    "technical-debt"
  ]
}

If you maintain this plugin, link to this source-backed listing from your README so users can review its manifest and indexed components.

[anti-hall on Agent Plugins Marketplace](https://pluginsmp.com/plugins/anti-hall)