skill-improve

Simulate feedback from census-grounded virtual populations. Panel-score product/copy, predict votes with IPF post-stratification, or run what-if social-sandbox experiments. Ships with 4 skills (core + product-feedback / vote-predict / social-sandbox), a bias audit module, and 6 eval suites (GSS, BFI, test-retest, diversity, demo-correlation, bias-audit).

4 skills

Adaptive add-on policy and recovery layer that decides WHEN a coding agent should compact, prune, snapshot, or fork -- replacing fixed-percent auto-compaction across Claude Code, Codex, and MCP-capable hosts. Deterministic 4-signal scorer (headroom/relevance/burden/continuity), zero-LLM pruning, state snapshots with secret redaction, host-capability-aware downgrade, self-installing hooks. Does not compact for you: it flags the moment + snapshots state + gates a triggered compaction; user runs /compact. No model calls, no network.

1 skill

Multi-agent, multi-device activity pool + per-project context layer. Captures Claude Code, Codex, Cursor, Maestri, and browser-use sessions across machines via a private GitHub repo, and distills recent activity into AGENTS.md + CLAUDE.md + docs/agent-context/ so any AI agent entering the project picks up where the last one left off.

1 skill

针对中国A股的行情与数据查询（基于本地 akshare 源代码与API映射）。当用户请求获取A股实时行情、历史K线（日/周/月）、分钟级分时（1/5/15/30/60）、复权数据（前复权/后复权）、股东户数（含详情）、以及巨潮资讯公告披露结果时使用本Skill。

1 skill

Capture a Mac app window as agent context — hover-pick a window, get a high-res screenshot plus Accessibility-tree text. Smart auto mode scroll-stitches scrollable windows into a seamless full-page image (sticky sidebars cropped, ratio-split tiles, silent, manual stop). Screenshot is primary context the agent reads with its own vision; AX text supplements. macOS only. No network, no model calls.

1 skill

Browser Use Cloud — 云端反检测浏览器，支持 Agent 模式（LLM 自动执行）和 Browser 模式（Playwright 直连，无 LLM 费用）。用于需要反检测、代理 IP、录屏截图、验证码绕过的自动化场景。

1 skill

Nightly offline self-evolution for this repo's Claude agent: harvests past Claude Code sessions (read-only), mines recurring tasks, replays them offline on your own API budget, and consolidates learnings into validated CLAUDE.md memory and SKILL.md skills behind a held-out gate, staged for review (never auto-applied). Verbatim vendor of microsoft/SkillOpt's stdlib-only skillopt_sleep engine + Claude Code plugin surface (MIT).

1 skill · 1 command · hooks

制作新 Skill，或改进已有 Skill。先按改动规模给出方案并等待确认，再生成或修改文件，并用真实任务对比改动前后的表现。用户要求做 Skill、把反复出现的问题沉淀成 Skill、修改或修复 Skill、根据失败案例改进 Skill、检查或测试 Skill 时使用；只有用户明确要求分享或发布时，才准备 GitHub 仓库并验证 `npx skills add`。

1 skill

Per-symbol maintenance optimizer for an agent's skill layer — learns skills from real runs, keeps them by adherence in use.

1 skill · 2 agents · 1 MCP server · hooks

Turns the corrections and repeated workflows of your work sessions into small, reviewed skill updates: every change comes with its evidence and a written reason, and nothing touches a skill until you approve and install it.

1 skill

Analytics engineering skills for warehouse exploration, dbt and semantic modeling, and schema-drift maintenance. Three skills (explore, transform, maintain) drive a local engine: warehouse reads are read-only and cost-gated, and project edits come back as reviewable diffs. Requires a local shell and uv; first use downloads the pinned engine from PyPI. No hooks, no MCP servers.

3 skills

Complete MaintainX integration skill pack with 24 skills covering work orders, preventive maintenance, asset management, and CMMS workflows. Flagship tier vendor pack.

24 skills