C

claude-code

Yet another software engineering agent orchestration system. This time with a mythical theme. Originally designed for Claude Code, now with support for OpenCode, Github Copilot, and OpenAI Codex.

Local agent-session record lake: Parquet + DuckDB over harness transcripts. Cost analytics, correction mining, simulation priors.

ask multivac_

multivac.ulm0.com

Before/after screenshots for GitLab UI merge requests: captured locally in a GDK checkout, verified, attached to the MR. Agent-agnostic (AGENTS.md); installable as a Claude Code plugin; no dependencies beyond the gitlab repo itself.

Ambiente di test per lavorare sui servizi digitali della PA con Linee Guida, Form.IO e AI

Specification graph manager for AI-augmented projects, on the nomograph-claim substrate.

GitLab's official AI skills vendored for GDK-free environments, plus a no-GDK contribution spine. Sibling of gitlab-simulation.

Ephemeral self-managed GitLab simulations on Google Compute Engine for LLM-mediated work.

MCP server for PostgreSQL. Single binary, no runtime dependencies.

One marketplace for every psyb0t agent skill and plugin, installable from Claude Code and Codex.

Download a ready-to-run AI-agent team you own. Works across Claude Code, Cursor, Codex and more, with no lock-in.

Describe an AI agent, get a ready-to-run agent home with persona, memory and a growth loop. Build one agent or a whole collaborating team.

Multi-agent collaboration and logging standard - pull rules, task handoff, flow-back, team journal (Claude Code / any LLM)

Let your AI agent reach your phone via ntfy - setup, CJK encoding pitfalls, when-to-call discipline (Claude Code / any LLM)

A predictions ledger with forced settlement for AI agents: log every judgment as a bet, settle it against reality, distill recurring biases into rules. Learns from outcomes, not corrections.

Know your user: a USER.md that grows. Your AI's understanding of you in a file you own — hand-written lane + agent-curated lane, no silent rewrites, moves when your folder moves.

Route each subagent to the cheapest capable model - about 70% off execution-tier tokens in our tests (Claude Code / any multi-tier agent platform)

Sharpen the axe before cutting wood - fool-proof requirement clarification skill for AI agents (Claude Code / Cursor / any LLM)

Teach your AI agent to acquire capabilities on its own - find existing skills before building one (Claude Code / any LLM)

A self-hosted AI platform — inference, tool use, browser automation, image generation, speech synthesis, transcription, object storage, agentic code execution, and more — behind a single OpenAI-compatible endpoint. One docker-compose up.