一键导入
evocli-memory
Use when you need to store or retrieve long-term information in an EvoCLI project. Covers constraint, episodic, semantic, and procedural memory types.
用 Codex 或 Claude 帮你安装 复制这段 Prompt,粘贴到 Codex、Claude 或其他助手里,让它检查 Skill 页面并帮你完成安装。
菜单
Use when you need to store or retrieve long-term information in an EvoCLI project. Covers constraint, episodic, semantic, and procedural memory types.
用 Codex 或 Claude 帮你安装 复制这段 Prompt,粘贴到 Codex、Claude 或其他助手里,让它检查 Skill 页面并帮你完成安装。
基于 SOC 职业分类
Use when connecting external MCP servers to EvoCLI or using EvoCLI as an MCP server for external AI clients.
Use when creating executable automation workflows (Skills) for an EvoCLI project. Covers TOML skill format, dry-run verification, lifecycle management.
Use this skill when working in an EvoCLI-powered project to leverage its unique capabilities: long-term memory, skills, workflow automation, and code intelligence.
| name | evocli-memory |
| description | Use when you need to store or retrieve long-term information in an EvoCLI project. Covers constraint, episodic, semantic, and procedural memory types. |
The EvoCLI memory system provides a persistent, tiered storage mechanism for project knowledge. It allows AI agents to maintain context across different sessions and even different users.
Knowledge is categorized by its scope and importance:
.evocli/memory/project/ directory. This contains information specific to the current repository. It is always loaded first.~/.evocli/memory/tools/. This contains learned patterns for specific tools (e.g., "The best way to run tests in this framework is...").~/.evocli/memory/global/. This contains your general preferences and cross-project knowledge.Constraints are the "laws" of the project. They are used to enforce coding standards, security policies, and architectural decisions.
any in TypeScript files."Episodic memory records "episodes" or events. It answers the question "What happened?".
Semantic memory stores facts and relationships. It answers the question "What is this?".
DataStore class is a singleton that manages the local SQLite connection."Procedural memory stores "how-to" guides.
Use the memory.add tool. You must specify the type and priority.
{
"content": "Use Tailwind for all new components.",
"type": "constraint",
"priority": "P1"
}
Use memory.query. This performs a hybrid search (vector + keyword) across all tiers.
{
"query": "How do we handle authentication?"
}
Memory is not static. Use memory.update to refine existing knowledge or memory.archive to remove outdated information.
EvoCLI's "Soul" periodically reviews episodic memories to identify patterns. If a certain fix is applied multiple times, it may be promoted to a Constraint or a Procedural memory.