long-horizon plugins

by Keli Wen

Dev skills I use day to day in agentic engineering, focused on multi-agent-driven work modes and long-horizon autonomous tasks. Includes long-horizon, fan-out, handoff, ctx-research, ctx-grill, to-ctx, just-ask, and the lowband and readout modes.

by MissingPackage

A harness for long-horizon Claude Code work: one map per project with the objective and the distance to it, a loop that kills ideas with cheap tests, hooks that catch the bad session, and four kinds of decision that stay yours.

by Bilal Khan

Multi-agent software delivery for Claude: discovery (brief, roadmap, business case, use cases), architecture, sprint execution with a plan→build→independent-QA triad, product-integrity traceability with drift detection and cross-sprint regression, live Cypress E2E, git workflow, and context discipline (measured facts, contract pre-flight, budgeted context packs). The state machine lives in scripts, not the model's memory.

by YK

Long-horizon task management for Claude Code — intake alignment, automated orchestration, independent verification

by Emmett Shaughnessy

Emmett Shaughnessy's personal Codex skills bundle. Provides SHAUGHV skills for self-routing long-horizon agentic prompt engineering and cross-runtime subagent brief compilation across software, data, mathematics, and science; falsifiable contracts, evidence receipts, Claude Fable 5 / Opus 5 and GPT-5.6 Sol / Codex overlays, prompt evaluation, loop escape, convergence recovery, evidence-aware retries, progressive slicing, critical thinking, OpenAI audio, PreText, Quiver AI, Mistral AI, brand design, animated brandmark, human changelog, CDN, naming, Google Cloud Storage, bug triage, design patterns, evidence-backed debugging, defensive programming, git workflow, security audit/red-team, typed session handoffs, iterative milestone planning, proportional workflow optimization, image generation/upscaling, deliberate learning, logical reasoning, personal productivity, optional MCP connection guidance, wb300, usage status line, and TT;DR. (The Claude-only subagent model/effort preference skill is omitted from Codex. The task + workplace-memory system lives in shaughv-tasks.)

by Emmett Shaughnessy

Emmett Shaughnessy's personal Codex skills bundle. Provides SHAUGHV skills for self-routing long-horizon agentic prompt engineering and cross-runtime subagent brief compilation across software, data, mathematics, and science; falsifiable contracts, evidence receipts, Claude Fable 5 / Opus 5 and GPT-5.6 Sol / Codex overlays, prompt evaluation, loop escape, convergence recovery, evidence-aware retries, progressive slicing, critical thinking, OpenAI audio, PreText, Quiver AI, Mistral AI, brand design, animated brandmark, human changelog, CDN, naming, Google Cloud Storage, bug triage, design patterns, evidence-backed debugging, defensive programming, git workflow, security audit/red-team, typed session handoffs, iterative milestone planning, proportional workflow optimization, image generation/upscaling, deliberate learning, logical reasoning, personal productivity, optional MCP connection guidance, wb300, usage status line, and TT;DR. (The Claude-only subagent model/effort preference skill is omitted from Codex. The task + workplace-memory system lives in shaughv-tasks.)