Autonomous ML research evolution engine: 11 specialized agents (Sage/Mutagen/Probe/Selector/Forge + sub-teams), MCP tools, a hook-enforced capability governor, a signal telemetry bus, and integrity/anti-gaming gates. The MCP server bundle ships prebuilt; the Python harness needs its deps (run ./install.sh once, or pip install -e harness).
Plans every prompt into a spec + Graph of Thought with HITL clarifications on a pluggable engine (Claude default, Grok optional), tracks it in Linear/Notion, and ships finished work through a Greptile-gated PR.
Expectation Architecture for all delivery modes. Routes context to strategy, keynote, or training mode. Maps expectations, detects mode drift, designs deliverables, and evaluates sessions.
Manage the Carta CRM conversationally — search, add, update, and enrich investors, companies, contacts, deals, notes, and fundraisings via the Carta CRM MCP Server.
Manage the Carta CRM conversationally — search, add, update, and enrich investors, companies, contacts, deals, notes, and fundraisings via the Carta CRM MCP Server.
The v2-only Bland plugin: migrate v1 pathways/personas to v2 agents (/norm:migrate), audit snapshots (/norm:validate), and verify with simulations and test chats (/norm:simulate). Raw-JSON workflow with the full hand-migration doctrine and trap catalog; no v1 authoring doctrine.
25 skills for taking a small web app from idea to shipped: sprint planning, scaffolding, refinement, React and Lighthouse optimization, a security pass, SEO indexing, on-demand MCP, delegation to external coding agents, token-budget checks, deploy, handoff, copywriting, documentation and legal policies. Ships the bld-executor subagent and a PreToolUse hook that blocks AI image generation.
Build, test, and run Bland voice agents from your coding agent: the hosted Bland MCP server, skills for every part of the platform, and the Norm agent builder.
24 agent skills for freelance software engineers — the full engagement lifecycle: vet, quote, contract, onboard, scope, build, deploy, communicate, bill, and close out.
Analyzes and optimizes CLAUDE.md files following Anthropic's official best practices. Use when reviewing existing CLAUDE.md for improvements, or when user mentions CLAUDE.md is too long or ineffective.
Installs skills from a user-curated manifest (~/.claude/skill-manifest.json) into the current project's .claude/skills/ — symlinks local skills, runs `npx skills add` for third-party ones, and advises `/plugin install` for native Claude plugins. Optionally scans the project for trigger files and pre-selects recommended matches. Ships a starter manifest as skill-manifest.example.json.