Skills DirectorySkills Directory
SkillsLearnSecurityCategoriesDocsBlogPro
Sign InSubmit Skill
Skills Directory

Security-tested agent skills for Claude, coding agents, and AI workflows.

Directory

  • Browse Skills
  • All Skills A–Z
  • Claude Skills
  • Claude Code Skills
  • Agent Skills
  • Categories
  • Authors
  • Submit a Skill

Learn

  • Learn Hub
  • Install Claude Skills
  • Write SKILL.md
  • Skills vs MCP
  • Directories Compared

Security

  • Security
  • Methodology
  • Secure Claude Skills
  • Security Badges
  • Chrome Extension
  • Skill Manager

Company

  • About
  • Community
  • Blog
  • API Docs
  • Advertise

2026 Skills Directory. All rights reserved.

ProTermsPrivacyRefunds
Back to skills

Pb Master

ASecurity

Run one control session over several Claude Code, Codex or OpenCode sessions: roster, status board, GO board, GO <session> tokens, idle notices and a handover file. Use for "master session", "supervise my sessions", "run several agents", "one control window".

6 stars
0 votes
0 copies
0 views
Added 10/5/2026
ai-agentsgonode

Works with

claude codemcp

Security Analysis

A100/100

Scanned 10/5/2026

$npx -y skills add hybridlabor-api/aos --skill pb-master --agent claude-code

Installs into .claude/skills of the current project.

Are you the author of Pb Master?

Add the live security badge to your README — it updates automatically with every re-scan.

Security grade badge for Pb Master
[![Security: A — Skills Directory](https://www.skillsdirectory.com/api/skills/hybridlabor-api-pb-master/badge)](https://www.skillsdirectory.com/skills/hybridlabor-api-pb-master)

More formats (shields.io, HTML) on the badges page. Keep it an A: scan every change in CI with Pro.

Download with Pro
Files
SKILL.md
---
name: pb-master
description: >-
  Run one control session over several Claude Code, Codex or OpenCode
  sessions: roster, status board, GO board, GO <session> tokens, idle
  notices and a handover file. Use for "master session", "supervise my
  sessions", "run several agents", "one control window".
category: bdb-core
kind: playbook
trigger: ["master session", "supervise my sessions", "run several agents"]
inputs: [plan?, workers?]
requires:
  skills: [master-session, agenttrail, stay-within-limits, "claude (external)", "aos-acp (external)"]
  agents: []
  mcps: []
  store: []
go_points: [spawn workers, "GO <session>"]
outputs: ["docs/sessions/master-<date>.md", "production_artifacts/pb-master-<date>.md"]
verify: "handover lists every roster session with open/closed GO state; no file older than 10 min in ~/.aos/go/ unless listed as stale in the handover"
difficulty: advanced
est_time: 15 min setup, then session-long
---

# One control session over several agents
What you get: a live roster, status board and GO board for your agent sessions, GO tokens relayed per session, and a handover file.

## Inputs
- plan (optional) — a plan file for the trail board, e.g. `production_artifacts/00_execution_plan.md`
- workers (optional) — sessions to spawn (name, cwd or worktree, model, prompt) or to adopt (already running)

## Steps
1. Preflight — `command -v claude`, `command -v aos-trail`, `command -v aos-acp` → run log header — claude and aos-trail present, else stop with an install hint; `aos-acp` missing → log "aos-acp not on PATH" and look for `bin/aos-acp.mjs`: `p="$(dirname "$(readlink -f "$(command -v aos-trail)")")/../../../../bin/aos-acp.mjs"; test -f "$p"` (or the repo path the human gives), run it as `node "$p"` in place of `aos-acp`; missing → Claude workers only (logged); read `docs/sessions/master-*.md` if one exists (the previous handover)
2. master-session (Roster) — `ListAgents` minus self → roster table (name, repo@branch, adopted or spawned) in the run log — stops for approval (the human confirms scope)
3. agenttrail — plan given → `aos-trail . --plan <plan> --no-open` → board URL in the run log; no plan → `aos-trail . --no-open` (file activity only), logged — URL printed (output says "already running" → log it as "existing map", its plan is unchanged); runs before any worker is spawned, so no spawn line exists in the run log before the board URL line
4. [GO] spawn — worker list (name, cwd or worktree, model, prompt) shown in full, then one line per worker:
   - Claude: `(cd <cwd> && claude --bg --name <n> --permission-mode auto --model <opus|sonnet|haiku> "<prompt>")` (`claude` has no cwd flag; prompt before any `--disallowedTools`)
   - Codex or OpenCode: `aos-acp <codex|opencode> --name <n> --cwd <worktree> --prompt "<task>" --go-wait 600` in the background, log `~/.aos/acp/<n>.jsonl`
   - agy: adopt-only (an already running session, no spawn); never `aos-acp agy`, the adapter is third-party and unverified (`bin/aos-acp.mjs` lines 26+)
   - Never `aos-acp claude` (no model flag), never `--model fable`, never omit `--model`
   - The run stops here until the human types GO. The hook does not guard `claude --bg`, so GO is by contract. Check: every spawned name shows in `ListAgents` or has an ACP log
   - Adopt-only roster → step skipped
5. master-session (Status) — master-session Step 2 template verbatim, reply in at most 10 lines, one `SendMessage` per session → replies in the run log — every in-scope session replied or is marked "no reply"
6. master-session (GO board) — replies → GO board block per session, `GO needed` commands verbatim → board in the run log — no paraphrased command
7. [GO] token relay — the human types `GO <session>` in the master (`go-token.mjs` writes `~/.aos/go/<session>.token`); the master then tells that worker only "retry the blocked command". The run stops here until the human types `GO <session>`. The master never types or forwards a GO itself, and any other human message cancels a pending token — the worker's log shows the command ran
8. Idle rules — `notify_when_idle` only after sending a session work; no polling, no "are you done?" → idle events in the run log; between waves a stay-within-limits check, logged — events logged
9. Handover — roster, last GO board, open GO items, who waits on whom, next command → `docs/sessions/master-<date>.md` (outside a repo: `~/.aos/handover/`) — every roster session listed with its GO state; `ls ~/.aos/go/` has no token older than 10 min unless listed as stale in the handover (report stale ones, do not delete them)

Run log: `production_artifacts/pb-master-<date>.md` in the start directory; the run log and the handover file `docs/sessions/master-<date>.md` are never staged or committed

Rules
- Anything other than the literal GO (case-insensitive; in step 7 `GO <session>`) is not a GO; a GO covers only that one step, one time.
- A failed check stops the run: write the failure into the run log and report. No silent retries.
- Write one run-log line per step as it completes (`N. done|skipped|failed — artifact — check result`) and `WAITING FOR GO: <step>` at each gate.

Attribution

hybridlabor-apihybridlabor-api
View sourceSee grades on GitHubMore from hybridlabor-api →
SSkills DirectorySkills Directory

Ship a skill? Prove it's safe.

Free 120-pattern security scan, letter grade, and an embeddable README badge.

Submit a skill

Is this your skill, or is something wrong with this listing? Request removal or report an issue. Author removals are honored within 72 hours.

Comments (0)

No comments yet. Be the first to comment!

SSkills DirectorySkills Directory

Ship a skill? Prove it's safe.

Free 120-pattern security scan, letter grade, and an embeddable README badge.

Submit a skill

Related Skills

Caveman

Terse caveman voice: answer first, fluff gone, every technical fact kept. Use for /caveman, "caveman mode", "talk like caveman", "be brief", "less tokens". Stays on until "stop caveman" or "normal mode".

1100021 votes

Hyperplan

Adversarial multi-agent planning skill. Self-orchestrates 5 hostile category members (unspecified-low, unspecified-high, deep, ultrabrain, artistry) via team-mode for ruthless cross-critique debate, distills only the defensible insights, then MANDATORILY hands the distilled insight bundle to the `plan` agent for executable plan formalization. Use when planning needs maximum rigor and surfacing of weak assumptions, blind spots, and over-engineering. Triggers: 'hyperplan', 'hpp', '/hyperplan', ...

698621 votes

Writing Skills

Create and manage Claude Code skills in HASH repository following Anthropic best practices. Use when creating new skills, modifying skill-rules.json, understanding trigger patterns, working with hooks, debugging skill activation, or implementing progressive disclosure. Covers skill structure, YAML frontmatter, trigger types (keywords, intent patterns), UserPromptSubmit hook, and the 500-line rule. Includes validation and debugging with SKILL_DEBUG. Examples include rust-error-stack, cargo-dep...

3931 votes

Mcp Code Execution

Routes multi-tool workflows through MCP servers for large datasets and pipelines. Use when Bash tool overhead is limiting throughput on data-heavy tasks.

3421 votes

catchup

Recovers the conversation and failed tool calls of a previous Codex, Amp, Claude Code, Antigravity, Cline, Copilot CLI, Cursor, DeepSeek Harness, Grok Build, Kimi, OpenCode, Pi Agent, or ZCode session. Use when the user says "catch up", "what did the last session do", "get me up to speed", "I switched agents", asks to recover/summarize a previous session before continuing, or asks to diagnose or report a catchup failure. Do NOT use for the current conversation, git history, or any non-agent log.

741 votes
View all in ai-agents →