inkwell-agent-sandboxes-and-software-factory

Name: inkwell-agent-sandboxes-and-software-factory
Rating: 74 (1 reviews)
Author: disler

disler/inkwell-agent-sandboxes-and-software-factory

Drive the herdr terminal-native agent multiplexer using natural language prompts to manage persistent fleets of terminal panes and AI coding agents.

97 24MITUpdated 2026-08-09

Good

View on GitHub Open source file

Overview

This skill provides comprehensive instructions for controlling herdr, a terminal-native multiplexer and socket server designed for tracking and orchestrating persistent terminal sessions and AI agents. It details how to programmatically manage workspaces, tabs, and panes, inspect semantic agent states, launch various coding agents like Claude Code or Codex, and automate complex multi-agent workflows without manual screen scraping.

Capabilities

▸Terminal pane and workspace orchestration
▸Semantic agent state tracking and blocking waits
▸Environment variable injection and git worktree management
▸Automated launch and control of CLI AI agents (Claude Code, Codex, pi, opencode)

Best for

Orchestrating multi-agent software development workflows inside terminal panes, Monitoring semantic agent states (idle, working, blocked, done) without polling UI screens, Running automated testing and build pipelines across isolated git worktrees, Managing persistent remote or local development sessions via CLI and JSON output

Works with

Claude CodeCodexCursorOpenCodeGemini CLI

Evoa Score breakdown= Σ (score × weight)

Task usefulness20%77 → +15.4

実タスクにどれだけ役立つか（機能の豊富さ・用途の明確さ）。 — AIによるcapabilities/use-cases解析

Code quality15%85 → +12.8

実装・指示の品質。 — AIによるSKILL.md/README解析

Maintenance15%100 → +15.0

リポジトリがどれだけ活発に保守されているか。 — GitHub 最終push日時の新しさ

Documentation12%41 → +4.9

ドキュメントの充実度・分かりやすさ。 — README/独自要約の情報量

Security15%50 → +7.5

危険・不審な挙動が無いか。 — AIによるセキュリティレビュー

Originality10%90 → +9.0

ありふれたラッパーではない独自性。 — AIによる独自性判定

Popularity8%54 → +4.3

コミュニティの採用度。 — GitHub Stars/Forks（対数スケール）

Compatibility5%100 → +5.0

対応AIエージェントの広さ。 — AIによる対応エージェント判定

Weighted total73.9 / 100

ライセンス不明/制限あり（Red）のSkillは総合スコアに0.85倍の補正を適用します。ランキングはこのScoreのみで決まり、広告で変わりません。算出方法の詳細 →

Security considerations

Instructions recommend using flags that bypass safety checks and permissions (e.g., --dangerously-skip-permissions for Claude Code) to facilitate unattended fleet automation, which requires careful scoping of tasks.