Releases·Tool Release
Sticky ruling plugin turns human decisions into subagent jobs so Claude Code keeps working while awaiting your ruling
A developer released sticky-ruling, a Claude Code plugin/skill that routes human decisions to an always-on-top window instead of freezing the session. Claude dispatches the ruling like a background subagent job and keeps working on tasks that don't depend on the answer. It's MIT-licensed, installable via the Claude Code plugin marketplace or a git-cloned skill, and tested on Windows only.
1 engagement1 sourceThu, Sep 24, 2026, 08:22 PM
AI Overview
- The problem: built-in question menus pin Claude Code questions but freeze the whole session until answered, so stepping away costs that work time.
- Sticky rulings dispatch the human's decision like a background subagent job; Claude keeps working on non-dependent tasks until the answer returns.
- The ruling appears in a small always-on-top window with full context, closable by clicking or replying in the terminal like "ruling 12: B".
- Install via /plugin marketplace add DrEvil-TitaniumHelix/sticky-ruling or as a plain skill cloned to ~/.claude/skills/sticky-ruling; MIT licensed.
- Tested on Windows only; the window never steals keyboard focus and works in any terminal or IDE.
Related
Social·Community
Developers discuss persistent project memory for coding agents
Developers using Claude Code and Cursor report that coding agents repeatedly re-ask or rediscover project decisions, wasting time and repeating mistakes. A developer proposes a CLI tool for persistent project memory, and the community discusses the problem's prevalence and potential solutions.
Developers using Claude Code and Cursor report that coding agents repeatedly re-ask or rediscover project decisions, wasting time and repeating mistakes. A developer proposes a CLI tool for persistent project memory, and the community discusses the problem's prevalence and potential solutions.
31091 engagement4 sourcesredditFri, Aug 14, 2026, 03:20 PM
Products·launched
Reasoning Doctrine: Agent Skill to Keep Goal in Focus
Reasoning Doctrine is a skill for AI agents that keeps the original goal in front of the agent throughout a long session, preventing it from losing track. It solves the problem of agents drifting from their objective during extended tasks. The skill is available on GitHub for integration into agent workflows.
Reasoning Doctrine is a skill for AI agents that keeps the original goal in front of the agent throughout a long session, preventing it from losing track. It solves the problem of agents drifting from their objective during extended tasks. The skill is available on GitHub for integration into agent workflows.
3 engagement1 sourceredditThu, Aug 27, 2026, 05:35 PM
Products·launched
Airport — VS Code extension for tracking multiple AI coding agents (Claude Code, Codex, Devin)
Airport is a VS Code extension that gives a unified view of parallel AI coding agents running in terminals, showing which are working, done, stuck, or awaiting permission prompts. It addresses the problem of alt-tabbing between identical-looking terminals to check agent status and losing track of which files each agent touches. Aimed at developers running several agents like Claude Code, Codex, or Devin at once.
Airport is a VS Code extension that gives a unified view of parallel AI coding agents running in terminals, showing which are working, done, stuck, or awaiting permission prompts. It addresses the problem of alt-tabbing between identical-looking terminals to check agent status and losing track of which files each agent touches. Aimed at developers running several agents like Claude Code, Codex, or Devin at once.
6 engagement1 sourceredditMon, Sep 21, 2026, 05:33 PM
Products·launched
persistent-handoff: Claude Code plugin giving restart-surviving memory via a skill and SessionStart hook
A Claude Code plugin consisting of a skill and a SessionStart hook that maintains one handoff file per agent. The agent rewrites the file at milestones, and the hook reinjects it into context on every session start, /clear, or /compact. It solves the problem of long-running Claude Code agents losing context after machine restarts, for people running many 24/7 agent sessions.
A Claude Code plugin consisting of a skill and a SessionStart hook that maintains one handoff file per agent. The agent rewrites the file at milestones, and the hook reinjects it into context on every session start, /clear, or /compact. It solves the problem of long-running Claude Code agents losing context after machine restarts, for people running many 24/7 agent sessions.
14 engagement1 sourceredditMon, Aug 31, 2026, 01:11 PM
Products·building
seedeep: live viewer for Claude Code subagent runs across multiple models
seedeep is a monitoring tool that tails the session file Claude Code writes and reconstructs an in-progress turn in real time. It shows the live subagent tree, each subagent's model, context usage, and outputs. It solves the problem of Claude Code's terminal showing only the last tool call and a percentage, for developers running multi-subagent turns.
seedeep is a monitoring tool that tails the session file Claude Code writes and reconstructs an in-progress turn in real time. It shows the live subagent tree, each subagent's model, context usage, and outputs. It solves the problem of Claude Code's terminal showing only the last tool call and a percentage, for developers running multi-subagent turns.
11 engagement1 sourceredditSat, Aug 29, 2026, 02:09 PM