evomap
- 21 skills
- 0 followers
- 13 hours ago last updated
- ▌ Evolver · evomapUse Evolver, the official EvoMap GEP-powered self-evolution engine, from Codex Desktop. Trigger when the user asks about Evolver, EvoMap, GEP, Genes, Capsules, EvolutionEvents, agent self-evolution, the Evolver Proxy, installing @evomap/evolver, running evolver in a git workspace, searching reusable evolution assets, publishing assets, or setting up Codex hooks for Evolver.
- ▌ Green Tea Perspective · evomap bundle绿茶的沟通操作系统。不是角色扮演教程,是可运行的社交策略与情感博弈框架。 基于依恋理论、社会心理学、间歇性强化研究、亲密关系动力学等 20+ 学术与文化来源的深度调研, 提炼 5 个核心心智模型、7 条决策启发式和完整的表达 DNA。 用途:以「绿茶」视角分析社交场景、情感博弈、人际操控与反操控策略。 当用户提到「绿茶模式」「绿茶视角」「用绿茶的方式分析」「green tea」时使用。 即使用户只是说「帮我用绿茶的角度看看」「如果绿茶会怎么做」「切换到绿茶」也应触发。
- ▌ Pdf2gep · evomap bundleConverts a PDF (local path or URL) into schema-valid GEP (Genome Evolution Protocol) assets for EvoMap. Emits a retrieval-oriented Gene (category=explore) plus a reference Capsule (source_type=reference) with a real asset_id. Does NOT produce execution Capsules; use skill2gep for that.
- ▌ Skill2gep · evomap bundleGeneric skill distillation tool. Converts any procedural Skill document (Cursor skill, Claude/Anthropic skill, or any SKILL.md / workflow-style markdown) into two kinds of GEP (Gene Evolution Protocol) assets. Gene = compact strategy template (signals + strategy + AVOID + validation). Capsule = audit record of a real execution of a Gene (outcome + execution_trace + env_fingerprint). Capsule must be produced from "Gene + at least one real-world execution"; fabricating a Capsule from the document alone is forbidden. Use when the user says "turn this skill into genes", "skill2gep", "skill2gene", "skill2capsule", "distill skill into GEP assets", or asks to compress a SKILL.md / procedural document into GEP control signals.
- ▌ Ar Coordinator · evomapAutoResearch 协调器:读取 idea 文件,在持久化 project_root 上调度 planner/coder/reviewer/runner。支持普通线性流水线,也支持 Ralph loop 驱动的可恢复工作流:每轮读取 state、做一个未完成单元、写回 state,全部完成时输出 <promise>AUTORESEARCH_DONE</promise>。Args = idea 文件路径 [可选 project_root]。
- ▌ Ar Gpu Preflight · evomapGPU need analysis, sizing, allocation, and pre-run RED/YELLOW/GREEN check on a Linux GPU server. MUST be used before running any code that imports torch/jax/tensorflow, mentions cuda/GPU, uses .cuda() / .to("cuda") / device="cuda", may allocate GPU memory, runs longer than 30s, or trains/infers a model. Triggers like nvidia-smi, CUDA, torch.cuda, GPU, 显存, 显卡, 跑代码, 运行代码, 跑一下, 训练, train, inference, smoke test, preflight, RED YELLOW GREEN, CUDA_VISIBLE_DEVICES, OOM. Independent of ar-experiment-runner — load and apply this skill even when the user gives a one-line GPU code request.
- ▌ Ar Workspace Safety · evomapHard filesystem and Python-environment safety rules for ar-runtime skills on a Linux GPU server. MUST be applied for ANY write, delete, move, install, or python invocation. Triggers like write file, save, mkdir, rm, delete, move, mv, cp, overwrite, cleanup, git reset, git clean, pip install, conda create, conda install, python3, python, miniconda, env, sudo, .ssh, .env, secret, credentials, /etc, /usr, /var, /opt, 删除, 安装, 写入, 清理. Independent of ar-experiment-runner — load and apply this skill even when no experiment context is in scope.
- ▌ Ar Experiment Runner · evomapar-runtime skills top-level experiment orchestrator on a Linux GPU server. MUST be used for any short experiment request such as "我需要做某个实验", "我要做一个实验", "帮我跑个实验", "做一下这个实验", "运行这个实验", "跑一下这段代码", "run an experiment", "create and run an experiment", "smoke test", "train a model", or any request to write/run experiment code. Coordinates profile inference, workspace creation, env setup, GPU preflight, execution, and artifacts in a single skill. Always co-applies ar-gpu-preflight and ar-workspace-safety.
- ▌ Skill Craft · evomap bundleCreate, review, and improve high-quality Agent Skills that follow proven engineering patterns. Use this skill whenever the user wants to create a new skill, improve an existing skill, review a skill for quality, convert a workflow into a reusable skill, or mentions "create skill", "write skill", "skill quality", "SKILL.md" — even if they just say "turn this into a skill" or "make this repeatable" (in any language).
- ▌ Chief Architect · evomapActivate the EvoMap Chief Architect persona for writing PRDs, architecture docs, naming concepts, and deep technical discussions. Use when the user asks to write project vision documents, seeks naming suggestions for concepts, requests high-level design docs, or engages in philosophical discussions about code architecture.
- ▌ Doc Source Sync · evomapDetect documentation drift when source code changes, and synchronize Wiki docs accordingly. Activate when user says "同步文档", "检查文档", "文档过期了吗", "doc drift", "sync docs", "update docs from code" — or when the user mentions that frontend code has changed and documentation might be stale. Also activate proactively when reviewing a PR or commit that touches files referenced in any doc's source_files frontmatter.
- ▌ Source Of Truth · evomapThe single gateway to all authoritative EvoMap knowledge: two source code repos and the official Wiki API. Use when the user says "获取前端", "前端代码", "查前端", "看前端", "website代码" to clone EvoMap/evomap-website, or "获取后端", "后端代码", "查后端", "看后端", "hub代码" to clone EvoMap/evomap-hub. Also activate when the user asks about EvoMap platform implementation details like "这个接口怎么实现的", "这个页面怎么写的", or references any EvoMap repo by name — even if they just say "帮我查一下代码" in the context of EvoMap development (in any language). Also activate when the user asks about EvoMap concepts, architecture, protocols, or any domain knowledge — e.g. "什么是 Evolution Capsule", "wiki", "文档", "Gene 是什么", "A2A 协议".
- ▌
- ▌ Evolver Run · evomapRun one Evolver self-evolution cycle on the current repo (collect signals → select/mutate genes → propose changes). Use when the user asks to "run evolution cycle", "start evolver loop", or "run evolver strategy".
- ▌ Evolver Sync · evomapSync evolution assets (genes/capsules) between the local store and the EvoMap Hub. Use when the user asks to "sync evolver", "publish gene", or "sync with hub".
- ▌ Evolver Evolve · evomapRun an evolution checkpoint — recall relevant past outcomes, reflect on the current task, and record what was learned. Use when the user requests an evolution checkpoint or asks to "evolve" or "checkpoint".
- ▌ Evolver Review · evomapReview Evolver's pending evolved changes, then approve (solidify) or reject (roll back). Use when the user asks to "review changes", "accept evolver changes", or "discard/rollback evolver changes".
- ▌ Evolver Status · evomapShow Evolver health — evolution memory location & size, recent log, and whether the full engine is installed. Use when the user asks for "evolver status", "evolver health", or "check evolver".
- ▌ Evolver Distill · evomapDistill a reusable skill/gene from recent run history (optionally from an LLM response file). Use when the user asks to "distill skill", "extract gene", or "generalize lessons".
- ▌ Evolver Solidify · evomapSolidify the current working changes into a durable Evolver gene/capsule (with rollback safety). Use when the user requests to "solidify changes", "save gene", or "capture this capsule".
- ▌ Capability Evolver · evomapSelf-evolution workflow for the agent. Before a substantive task, recall what worked on similar past tasks from evolution memory; after it, record the outcome so future sessions learn from it. Use when the user starts non-trivial work (a feature, a fix, a refactor) or asks the agent to "evolve", "learn from this", or "remember how this went".