PickySteve
Skill router and context picker for coding agents: a local model (via Ollama, offline by default) routes each prompt to the right skill via hybrid BM25 + embedding retrieval, cross-encoder rerank, and an LLM judge, then assembles a minimal context bundle for the execution…
- from
- Gemini CLI
- added
- 2026-10-10
- likes
- 0
similar
-
SkillAgent skillagent.dev
Remote MCP server (Streamable HTTP, no auth) for searching ~3,500 agent skills, rules files (including Gemini rules) and MCP servers indexed hourly from GitHub, with project-based recommendations and per-agent install instructions for Gemini CLI. Install: gemini mcp add…
-
keen-code github.com
A context-efficient terminal-based AI coding agent. Provider agnostic, supports MCPs, Agent Skills, Subagents, and more. Comes with a simple and straightforward TUI.
-
Emdash Skills github.com
14-category autonomous product-building OS with 18 specialized agents. Turns one-line prompts into deployed products. Includes GEMINI.md compatibility for cross-tool portability. Skills cover architecture, planning, quality, brand, media, observability, and deployment on…
-
claude-skills-pro github.com
15 engineering-workflow skills (7-dimension code review, root-cause debugging, bug-catching test generation, behavior-preserving refactor, zero-downtime DB migration; 7 free/MIT) with CN/EN handbooks. Installs into Gemini CLI via the open skills installer: npx skills add…
-
Pluribus github.com
Sync one canonical project context into native AI coding-agent rule files, including Gemini CLI GEMINI.md, Claude Code CLAUDE.md/AGENTS.md, Cursor, Cline, Roo Code, Amazon Q, Junie, Warp, Copilot, Windsurf, Continue, Zed, OpenCode, and OpenClaw.
-
Bernstein github.com
Multi-agent orchestrator that coordinates Gemini CLI alongside Claude Code and Codex CLI. Spawns parallel coding agents from a single goal, verifies with tests, auto-commits. Deterministic Python coordination, zero LLM tokens on orchestration.
Gemini CLI › Agent Orchestration & CLI Tools: “Skill router and context picker for coding agents: a local model (via Ollama, offline by default) routes each prompt to the right skill via hybrid BM25 + embedding retrieval, cross-encoder rerank, and an LLM judge, then assembles a minimal context bundle for the execution…”