Lightweight standalone ControlFlow plugin for Claude Code: high-quality plan generation in the shared ControlFlow format, inline adversarial plan verification (zero subagents), and evidence-backed review with proactive vulnerability/error search layered over the native Claude Code toolset. 3 skills, 0 subagents …
Durable goal-following for Claude Code: contracts with definition-of-done, subagent judge gates, executor-subagent chain execution, and adaptive missions — with all state mechanics executed by a tested CLI and invariants enforced by hooks. Inspired by OpenAI Codex /goal and the Ralph loop pattern.
LLM-maintained personal wiki. Claude actively ingests, synthesizes, and cross-links knowledge across all your projects. Based on Karpathy's LLM Wiki pattern.
★not rated 13 4mo agoA
tokens not measured
originalMIT
Multi-agent orchestration plugin - a virtual dev crew with intelligent model routing (Opus/Sonnet/Haiku), lite mode, 3-layer codebase indexing, auto-skill selection, task history, rollback, and custom agents.
★not rated 13 5mo agoA
tokens not measured
originalMIT
Spec-driven agentic engineering as graph engineering: shape (pitch) -> decide (ADR) -> execute (delegated cycle with adversarial review) -> measure -> eval, with /graph to fan work out across a fleet of subagents (diamond / router / verifier / cycle). Dark-launch eval-gated. Distilled from a real practice.
★not rated 13 1mo agoA
tokens not measured
originalMIT
Advisor Loop — an autonomous loop for long, multi-day work: an independent advisor certifies mission completion and the anchor survives every compaction, on subscription-safe agents.
★not rated 13
changed 9d agoA
tokens not measured
originalApache-2.0
Orchestrator plugin that weaves planning-with-files, DESIGN.md, memory routing, agent teams, and session handoff into one composable contract for any modern coding agent. Composes existing skills without altering their core. Includes code-to-DESIGN.md harvest and critique.
★not rated 13 4mo agoA
tokens not measured
originalApache-2.0
Agent memory, charged rent. A Stop hook records what each session costs; expensive runs are distilled into candidate efficiency rules; each candidate is benchmarked on a frozen golden suite and kept only if it saves at least twice its context rent. Six commands. Two theorems survived measurement and three were deleted.
★not rated 13
changed 3d agoA
tokens not measured
originalMIT
Stop your AI agent from generating Tailwind CSS v3 code. 20+ rules for v4 syntax, CSS-first config, modern utility patterns, and common anti-patterns agents produce.
★not rated 13 6mo agoA
tokens not measured
originalMIT
Audit any Chrome extension against Chrome Web Store Featured-badge and Established Publisher badge criteria, and produce a prioritized fix plan to qualify.
★not rated 13 2mo agoA
tokens not measured
originalMIT
Architecture, code review, prompting, and decision-making skills tuned to Alon Baron's engineering style: model-first, failure-path-aware, YAGNI. v2: skills self-invoke when the task shape matches. v2.1: triggers name the situation and settle overlaps with Claude Code's built-ins.
★not rated 13 1mo agoA
tokens not measured
originalMIT
At most 3 mods per repository are shown here, and a mod shipped inside a plugin is left to that plugin's page — the rest are on their repository pages: