Stats
Links
Categories
npx claudepluginhub kennythelema/prototype_agentSuperpowers for Claude Code — dev tools, ethical career agent, and hooks that turbocharge your workflow
An AI career agent built on Claude Code that helps people find meaningful work — with honesty, values alignment, and genuine human connection at the center.
11 skills that chain together into a full career pipeline:
| Skill | Command | What It Does |
|---|---|---|
| Profile Setup | /profile-setup | Build your career profile — values, deal-breakers, preferences |
| Job Search | /job-search | Curated, scored results with ethical company signals |
| Evaluate Company | /evaluate-company | 7-dimension ethical rubric (35-point scale) |
| Tailor Resume | /tailor-resume | Honest tailoring — highlights real fit, flags real gaps |
| Cover Letter | /cover-letter | Specific, authentic letters (refuses to write generic ones) |
| Interview Prep | /interview-prep | STAR stories, company research, questions to ask back |
| Application Tracker | /application-tracker | Track stages, follow-ups, and conversion |
| Community Find | /community-find | Discover communities aligned with your craft and values |
| Connect Request | /connect-request | Genuine outreach (enforces the Genuine Connection Test) |
| Side Project Scope | /side-project-scope | Scope ideas into weekend, quarterly, or full venture |
| Find Collaborators | /find-collaborators | Find co-founders from communities and directories |
It has opinions. Job search results are scored and ranked, not dumped as a list. Company evaluations flag real concerns. Cover letters are refused when there's no genuine connection to build on.
It refuses to help you lie. Won't fabricate resume credentials, generate mass-application spam, or draft dishonest outreach. Surfaces gaps honestly so you can address them — in the cover letter, in the interview, or by finding a better-matched role.
It centers values. Your deal-breakers are first-class filters. If a company triggers one, the agent leads with that — before any other analysis.
Built as a Claude Code agent with custom skills, hooks, and test personas:
.claude/
├── agents/ethical-career-agent.md # Unified agent definition
├── skills/ # 11 skill definitions
├── data/
│ ├── personas/ # 6 test personas for validation
│ └── interactions/ # 11 example interactions (good + bad)
├── docs/ # Architecture docs and data schema
└── settings.json # Hooks and permissions
User data stays local at ~/.career/. Nothing is sent externally.
The interaction logs in .claude/data/interactions/ document both beneficial and unhelpful agent behaviors. The unhelpful examples — preachy lecturing, dismissive shaming, resume fabrication, uncritical cheerleading — directly informed the Critical Rules in every skill definition.
6 personas with detailed backstories, emotional triggers, and success criteria:
/profile-setup to build your career profile/job-search [role] to start searchingClaude Code — Anthropic's CLI and IDE extension for Claude.
Open Design — local-first design app exposed to coding agents over MCP. Install once with your agent's plugin command and projects/files/skills are reachable through stdio.
Claude Code plugins for the Slidev presentation framework
Bundled plugins for actuating and debugging the Chrome browser.