eld-model-law-discovery

eld-model-term-card

>

AGI 环境一键同步工具。从 GitHub 仓库同步所有配置到本地：Skills 技能库、全局提示词、MCP 服务器配置、Output Styles。支持 Claude Code、Codex CLI、OpenClaw、Pi Coding Agent 等多平台。适用于多设备统一环境、换电脑恢复、团队共享配置等场景。

Propagator-based parallel structure for exploiting knowledge differentials

Erlang-style message passing via /tmp file system. Use when simulating actor model with actual message synchronization between processes. Provides send/receive primitives with blocking and pattern matching.

Estimate implementation complexity for features and technical tasks

Estimation techniques including analogous, parametric, three-point, and expert judgment methods. Provides effort, cost, duration, and complexity estimates for projects, features, and tasks.

>-

exploration-continue

**Parallelize exploration aggressively.** When researching or exploring, launch multiple subagents simultaneously instead of sequential searches. This reduces wall-clock time proportionally to the num

faber-director-legacy

The `block-commands` skill documents how to use `.claude/settings.json` deny rules to block dangerous commands. However, `.claude/settings.json` is a repository-local file that is not distributed with

feature-documenter

Automate the complete lifecycle of feature flags in Saberloop, following Epic 10 hygiene standards

**创建时间**: 2026-01-25

dev:feedback

**Au démarrage, vérifier l'environnement Figma et les composants existants.**

Stop asking "How long will this take?" and start asking "How much is this worth to us?" By setting a fixed budget (an appetite) rather than a variable estimate, you force teams to find the simplest, m

Manage Flipt feature flags - list, create, enable/disable, and configure rollout rules. Use when you need to control feature flag state or set up segmented rollouts.

Create Forgekeeper task cards following project conventions

forward-derivation

foundations

First-time setup for the foundry plugin (plugin:foundry@claude-foundry)

The **GAM Researcher Agent** automates retrieval and synthesis of context from your Transmission Packet archive. It eliminates manual context pasting by implementing an iterative research loop that se

gas-clasp-ops

Analyze large codebases using Gemini CLI's massive context window. Use when analyzing entire codebases, comparing multiple large files, understanding project-wide patterns, verifying feature implementations, or when context limits are exceeded.

>

**Purpose:** Strategic enhancement and functional evolution of agent skills

Create, switch, and manage git branches with proper handling of uncommitted changes. Use when creating feature branches, switching contexts, or cleaning up after merges.

git-safety-guard

Use this skill in the scenario of writing git commit messages.

Creates well-structured GitHub issues for the MCPSpy project using the gh CLI tool. Use when asked to create issues, report bugs, or document features. Follows conventional naming with feat/chore/fix prefixes and maintains appropriate detail levels.

GitHub issue lifecycle management with worktree isolation

Spawn and coordinate parallel Claude Code sub-agents using git worktrees. Use this skill when you need to parallelize work across multiple independent tasks, delegate subtasks to sub-agents, or orchestrate complex multi-part implementations.

List and manage cross-session goals and milestones. Use for tracking multi-session work, viewing progress, or managing milestones.

Tracks micro-sprint completion state and progress for Google Sheets operations

GPU-aware PPO training configuration for A100/H100. Trigger when training is slow or GPU utilization is low.

Run diagnostic checks on GSD installation and report status

Use when starting a session (resume previous work) or ending a session (create handover for next). Automatically detects mode based on conversation state. Cascades through completed handovers via Haiku subagents until finding incomplete work.

View and analyze Hawk evaluation results. Use when the user wants to see eval-set results, check evaluation status, list samples, view transcripts, or analyze agent behavior from a completed evaluation run.

hcom

hecras_compute_plans

hecras_compute_remote

Claude Code Hook system reference for capabilities, possibilities, and limitations. Use when you want to know what hooks can do.

Claude Code Hook 开发专家。当需要创建、调试、配置 Hook 时触发。触发词：创建 hook、写 hook、hook 配置、stop hook、pre tool use、post tool use、session start。

>

<objective>

- 在Julia中运行

'Query and manage feature implementation task status. Provides CLI tools to list features, check task status, find ready tasks, and validate task files. Used by /implement-feature orchestrator to track progress. Automatically updates task timestamps via hooks on /start-task.'