cosmo

Nightly system janitor sweep. Cleans up workspace, tests infrastructure, fixes what it can, and sends a report to Sid's DM.

OpenAI Codex CLI wrapper — three modes. Code review: independent diff review via codex review with pass/fail gate. Challenge: adversarial mode that tries to break your code. Consult: ask codex anything with session continuity for follow-ups. The "200 IQ autistic developer" second opinion. Use when asked to "codex review", "codex challenge", "ask codex", "second opinion", or "consult codex". (gstack) Voice triggers (speech-to-text aliases): "code x", "code ex", "get another opinion".

Set up gbrain for this coding agent: install the CLI, initialize a local PGLite or Supabase brain, register MCP, capture per-remote trust policy. One command from zero to "gbrain is running, and this agent can call it." Use when: "setup gbrain", "connect gbrain", "start gbrain", "install gbrain", "configure gbrain for this machine". (gstack)

Codify the most recent successful /scrape flow into a permanent browser-skill on disk. Future /scrape calls with the same intent run the codified script in ~200ms instead of re-driving the page. Walks back through the conversation, synthesizes script.ts + script.test.ts + fixture, runs the test in a temp dir, and asks before committing. Use when asked to "skillify", "codify", "save this scrape", or "make this permanent". (gstack)

Keep gbrain current with this repo's code and refresh agent search guidance in CLAUDE.md. Wraps the gstack-gbrain-sync orchestrator with state probing, native code-surface registration, capability checks, and a verdict block. Re-runnable, idempotent. Use when: "sync gbrain", "refresh gbrain", "re-index this repo", "gbrain search isn't finding things". (gstack)

Auto Review closeout. Codex review is the default when no engine is set and is the recommended reviewer.

CEO/founder-mode plan review. Rethink the problem, find the 10-star product, challenge premises, expand scope when it creates a better product. Four modes: SCOPE EXPANSION (dream big), SELECTIVE EXPANSION (hold scope + cherry-pick), HOLD SCOPE (maximum rigor), SCOPE REDUCTION (strip to essentials). Use when asked to review a plan, challenge this, CEO review, poke holes, think bigger, or expand scope.

Systematic debugging with root cause investigation. Four phases: investigate, analyze, hypothesize, implement. Iron Law: no fixes without root cause. Use when asked to debug, fix a bug, investigate an error, or root cause analysis. Proactively use when user reports errors, stack traces, unexpected behavior, or says something stopped working.

Product interrogation with six forcing questions. Two modes: startup diagnostic (demand reality, status quo, desperate specificity, narrowest wedge, observation, future-fit) and builder brainstorm. Use when asked to brainstorm, "is this worth building", "I have an idea", "office hours", or "help me think through this". Proactively use when user describes a new product idea or wants to think through design decisions before any code is written.

Weekly engineering retrospective. Analyzes commit history, work patterns, and code quality metrics with persistent history and trend tracking. Team-aware with per-person contributions, praise, and growth areas. Use when asked for weekly retro, what shipped this week, or engineering retrospective.

世界顶级思维合集 —— 融合Google Staff Engineer、Martin Fowler/Kent Beck/Jeff Dean工程思维、Paul Graham/Sam Altman创业思维、Elon Musk创新思维、Stripe/Airbnb设计思维。v2.5.10：移除install.sh以完全消除ClawHub Suspicious警告。

性能基准工程师 —— 像 Google PageSpeed 和 WebPageTest 一样进行性能测试，建立基准，追踪回归，优化 Core Web Vitals

浏览器测试工程师 —— 像 Playwright 团队、Chrome DevTools 工程师和 WebPageTest 一样进行专业的浏览器测试、性能分析和可视化回归测试。

金丝雀监控工程师 —— 像 Netflix 和 Google SRE 一样监控金丝雀发布，自动检测异常，智能决策回滚或继续

DevOps工程师 —— 像 Kelsey Hightower（Kubernetes布道者）、Brendan Burns（K8s创始人）和 Thomas Ptacek（安全专家）一样设计部署架构。融合云原生、GitOps和安全最佳实践。

高级设计评审师 —— 像 Stripe、Airbnb 的设计团队一样评审设计，检测 AI Slop，生成完整设计系统，打造 10-Star 体验

技术文档工程师 —— 像 Stripe、Dropbox 和 MDN 团队一样编写世界级文档。融合技术写作最佳实践，自动生成清晰、专业、用户友好的项目文档。

GitHub集成助手 —— 像 GitHub Actions、Mergify 和 Semantic Release 一样自动化 PR 管理、CI 监控和发布流程。

项目初始化助手 —— 像 Facebook/Meta 的 Create React App、Vercel 的模板系统和 Cookiecutter 项目一样，标准化、自动化地创建项目骨架。

系统调试专家 —— 像 Netflix 的 SRE 和 Google 的 Debugging 团队一样进行系统性根因分析，3次失败自动停止，科学定位 Bug

部署验证工程师 —— 合并PR、部署到生产、验证健康状态，一键完成从代码到生产的全流程

消息通知助手 —— 像 PagerDuty、OpsGenie 和 Slack 的现代通知系统一样，智能路由、分级告警、多渠道通知。

YC 合伙人级别的创业导师 —— 像 Paul Graham 和 Sam Altman 一样问尖锐问题，找到真正的创业机会

YC CEO / 产品经理模式 —— 融合 Brian Chesky、Paul Graham 的产品思维 + Elon Musk 的第一性原理、长期主义、规模化思维

像 Martin Fowler、Kent Beck、Jeff Dean 一样设计系统 —— 简单优雅、可演进、高可用的架构

Google SET/测试架构师级别的质量保障 —— 像Google一样做测试，构建高质量软件

复盘师 —— 像 Spotify 的 Squad Health Check、Google 的 Project Aristotle 和 Atlassian 的团队健康度实践一样进行数据驱动的项目复盘和团队改进。

Google Staff Engineer 级别的代码审查 —— 像谷歌一样做CR，构建世界级代码质量

首席安全官 —— 像 Google Security Team 和 OWASP 专家一样进行安全审计，OWASP Top 10 + STRIDE 威胁建模，零误报安全扫描

发布工程师 —— 像 Jez Humble、Gene Kim 和 Charity Majors 一样专业发布。融合持续交付、SRE 和可观测性最佳实践，确保每次上线都稳定可靠。

项目状态追踪 —— 像 GitHub Projects、Linear 和 Jira 的项目管理最佳实践一样追踪项目进度、识别阻塞、推荐下一步行动。

测试工程师 —— 像 Kent Beck（TDD之父）、James Whittaker（Google测试架构师）和 Cem Kaner（软件测试之父）一样设计全面的测试策略。融合测试驱动开发、探索性测试和自动化最佳实践。

Generate self-contained, interactive HTML lead magnets, proposals, and GTM assets. Use when asked to create a lead magnet, one-pager, interactive proposal, competitive analysis page, onboarding doc, sales page, audit deliverable, or any single-file HTML asset designed to impress prospects before a sales call. Also triggers on "build me an HTML page for outbound", "create a pre-demo deliverable", "make a custom proposal page", or "generate a GTM asset". NOT for full websites, multi-page apps, or standard landing pages with backends.

LinkedIn growth system for founders, builders, and consultants. Use when planning LinkedIn content strategy, designing a content calendar, evaluating LinkedIn performance, or coaching on LinkedIn for pipeline and authority. Covers the full-funnel system, 2026 algorithm mechanics, format rankings, and case studies. Works for SaaS, consulting, agencies, research labs — anyone building in public and converting attention to business. For actually writing posts, use linkedin-writer. For scheduling/posting ops, use linkedin-posting.

Operational skill for LinkedIn content scheduling, posting cadence, calendar planning, engagement tactics, and growth strategy. Use when planning a content calendar, deciding when/how often to post, setting up amplification, managing warm outbound, or tracking performance metrics. For writing posts, use linkedin-writer. For strategy, use linkedin-saas-playbook.

Copywriting skill for generating LinkedIn posts on behalf of a founder. Use when writing, drafting, or revising any LinkedIn post. Handles hooks, formatting, tone, anti-slop, and quality gates. Works in concert with linkedin-saas-playbook (strategy) and linkedin-posting (scheduling/ops). References the copywriting skill's anti-slop rules and content guardrails.

Nightly value creation session. Review the goal, assess current state, identify the highest-leverage thing to build, ship it, and report to Sid. v2: Expanded scope — can build product features, improve plugins, create customer-facing tools, ship code via Claude Code, and run autoreview.

Self-improving learning loop. After conversations, review what happened and update skills and memory. Runs during heartbeats by reviewing recent session transcripts. Adapted from Hermes Agent's background_review pattern.

Full-text search across past conversations using SQLite FTS5. Three modes: DISCOVER (query search), SCROLL (anchor + window), BROWSE (recent sessions). Adapted from Hermes Agent's session_search_tool pattern.

Periodic maintenance of agent-created skills. Auto-transitions lifecycle states (active→stale→archived), consolidates overlapping skills into umbrellas, and maintains the skill collection's health. Adapted from Hermes Agent's curator pattern. Runs weekly during heartbeats.