Agent Plugins Marketplace
← All plugins

herdr-crew

v0.2.10

The orchestrator/worker session model over herdr. An orchestrator session keeps its context for decisions and dispatches everything else to worker sessions launched by seat name: the operator's own files name each seat (~/.claude/herdr-crew/agents.md on the machine, .herdr-crew/roles.md in the project), implementers route by complexity class, one reviewer session runs /code-review once per PR, the retained implementer works GitHub threads to zero, merge only on the operator's word. A session budget caps each work item at one implementer and one verifier. Each seat gets its own labelled tab holding one full-size pane, in the tree its --cwd names, so a run of five seats is five readable tabs. Readiness is derived, not declared: a seat herdr detects waits on a typed agent state, a seat it does not waits on a screen pattern, so an undetected agent CLI is a first-class seat and a detection manifest later upgrades it with no edit. A run's state lives in a dagr run.json the plugin owns and no binary writes. One fail-open hook, on Claude Code only, tells a seat its own context figure past a ceiling of 500000 tokens — a coordinator seat rotates on it, a leaf finishes; where molt runs, a child hands off to its parent and the top molts in place. Defers every other herdr command to the binary's own guide. Since 0.2.0, a spine's spine and close seats can be a DeepSeek Harness session (kind dsh-spine-driver, with dsh-crew), its items routed by .dsh-crew/roles.md. No lib, no state.

CodexClaude Code1 Skill

By PrasLicense: MIT0 GitHub starsUpdated 2 weeks ago

Directory evidence

Runtimes
Codex and Claude Code
Parsed components
1 skill or MCP entry
Source updated
Oct 4, 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 herdr-crew for Codex and Claude Code

Installs for the current user
codex plugin marketplace add pulseai-labs/claude-agent-scaffolding
codex plugin marketplace upgrade claude-agent-scaffolding-codex
codex plugin add herdr-crew@claude-agent-scaffolding-codex

Paste and run these commands in a terminal with Codex. They add and refresh the claude-agent-scaffolding-codex catalog, then install this plugin.

Compatibility: the page URL and API slug “herdr-crew” remain stable.

  • Codex: herdr-crew@agent-plugin-marketplace → herdr-crew@claude-agent-scaffolding-codex

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/pulseai-labs/claude-agent-scaffolding

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

Plugin files

herdr-crew/
├── .codex-plugin/plugin.json
├── .claude-plugin/plugin.json
└── skills/orchestrate/SKILL.md

Included Skills1

orchestrateskills/orchestrate/SKILL.md

The orchestrator/worker session model over herdr — one orchestrator session that spends its context on decisions and dispatches everything else to worker sessions launched by seat name through herdr, the seats defined in the operator's own files (~/.claude/herdr-crew/agents.md, .herdr-crew/roles.md). One /code-review per PR, findings returned in the seat's report file, GitHub threads worked to zero, merge only on the operator's word. On a spine this session just planned, the operator approves the seats into the project file and they are injected into one spine session that gives each item seat a fresh tab and pane. Use when the user says the herdr orchestrator session, a herdr worker seat, the herdr crew, dispatch to a herdr session, execution assignments for a spine in herdr, or runs /herdr-crew:orchestrate. Not herdr's command reference (`herdr --skill` owns that), and not a PR loop of its own where ossify's work-pr is installed.

Plugin manifests2

herdr-crew/.codex-plugin/plugin.json
{
  "name": "herdr-crew",
  "version": "0.2.10",
  "description": "The orchestrator/worker session model over herdr. An orchestrator session keeps its context for decisions and dispatches everything else to worker sessions launched by seat name: the operator's own files name each seat (~/.claude/herdr-crew/agents.md on the machine, .herdr-crew/roles.md in the project), implementers route by complexity class, one reviewer session runs /code-review once per PR, the retained implementer works GitHub threads to zero, merge only on the operator's word. A session budget caps each work item at one implementer and one verifier. Each seat gets its own labelled tab holding one full-size pane, in the tree its --cwd names, so a run of five seats is five readable tabs. Readiness is derived, not declared: a seat herdr detects waits on a typed agent state, a seat it does not waits on a screen pattern, so an undetected agent CLI is a first-class seat and a detection manifest later upgrades it with no edit. A run's state lives in a dagr run.json the plugin owns and no binary writes. One fail-open hook, on Claude Code only, tells a seat its own context figure past a ceiling of 500000 tokens — a coordinator seat rotates on it, a leaf finishes; where molt runs, a child hands off to its parent and the top molts in place. Defers every other herdr command to the binary's own guide. Since 0.2.0, a spine's spine and close seats can be a DeepSeek Harness session (kind dsh-spine-driver, with dsh-crew), its items routed by .dsh-crew/roles.md. No lib, no state.",
  "author": {
    "name": "Pras"
  },
  "repository": "https://github.com/pulseai-labs/claude-agent-scaffolding",
  "license": "MIT",
  "keywords": [
    "orchestration",
    "herdr",
    "multi-agent",
    "code-review",
    "skills"
  ],
  "skills": "./skills/",
  "interface": {
    "displayName": "Herdr Crew",
    "shortDescription": "Orchestrator/worker session model over herdr.",
    "longDescription": "Herdr Crew states who does what when one orchestrator session drives worker sessions through herdr: which named seat fills each role, what a brief must carry, where the orchestrator stops and dispatches instead, one code review per PR with findings returned through the seat's report file, GitHub threads worked to zero, and a merge only on the operator's explicit word. It defers every other herdr command to the herdr binary's own version-matched guide.",
    "developerName": "Pras",
    "category": "Coding",
    "capabilities": [
      "Interactive",
      "Read",
      "Write"
    ],
    "defaultPrompt": [
      "Start a herdr orchestrator session for this objective.",
      "Dispatch a herdr worker seat for this task.",
      "Review PR 123 in a fresh herdr seat."
    ],
    "websiteURL": "https://github.com/pulseai-labs/claude-agent-scaffolding/tree/main/herdr-crew",
    "brandColor": "#1D4ED8"
  }
}
herdr-crew/.claude-plugin/plugin.json
{
  "name": "herdr-crew",
  "version": "0.2.10",
  "description": "The orchestrator/worker session model over herdr. An orchestrator session keeps its context for decisions and dispatches everything else to worker sessions launched by seat name: the operator's own files name each seat (~/.claude/herdr-crew/agents.md on the machine, .herdr-crew/roles.md in the project), implementers route by complexity class, one reviewer session runs /code-review once per PR, the retained implementer works GitHub threads to zero, merge only on the operator's word. A session budget caps each work item at one implementer and one verifier. Each seat gets its own labelled tab holding one full-size pane, in the tree its --cwd names, so a run of five seats is five readable tabs. Readiness is derived, not declared: a seat herdr detects waits on a typed agent state, a seat it does not waits on a screen pattern, so an undetected agent CLI is a first-class seat and a detection manifest later upgrades it with no edit. A run's state lives in a dagr run.json the plugin owns and no binary writes. One fail-open hook tells a seat its own context figure past a configurable ceiling (default 500000 tokens) — a coordinator seat rotates on it, a leaf finishes; where molt runs, a child hands off to its parent and the top molts in place. Defers every other herdr command to the binary's own guide. Since 0.2.0, a spine's spine and close seats can be a DeepSeek Harness session (kind dsh-spine-driver, with dsh-crew), its items routed by .dsh-crew/roles.md. No lib, no state.",
  "author": {
    "name": "Pras"
  },
  "userConfig": {
    "context_ceiling": {
      "type": "number",
      "title": "Context ceiling (tokens)",
      "description": "Past this many tokens of context, a herdr-crew coordinator seat (the top, a spine session, a work-PR session) is told to finish the unit in hand, start no new one and rotate at its next boundary. 500000 is half a 1M window.",
      "default": 500000,
      "min": 1
    }
  }
}

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

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