Token导航 LogoToken导航TokenDH.com
研究检索敏感数据github未标认证来源可访问clear审计未展示

session-memory会话内存

Agent Skill

session-memory 用于查找、检索和筛选相关信息,适合在 Codex、Claude、Cursor、Gemini CLI 中需要根据关键词、任务场景或来源线索快速定位候选结果时使用。可结合来源仓库、安装命令和原始 README 继续核验具体用法。安装前建议确认权限范围、维护状态,以及是否会触发联网、命令执行或文件读写。

总安装

212

周安装

9

GitHub Stars

公开资料未说明

下载量

74
CodexClaudeCursorGemini CLI

安装说明

本站只整理中文说明和来源信息,不托管安装包,也不代用户安装。

GitHub

来源数

2

许可证

MIT

最后核验

2026-05-01

来源状态

来源可访问

安装方式

通过对话安装

复制提示词发给支持本地命令或 Skills 的 AI 助手,先确认命令和权限,再让它执行。

请帮我安装这个 Agent Skill:session-memory(会话内存)
来源仓库:https://github.com/anortham/goldfish
仓库路径:skills/session-memory
安装命令:
npx skills add anortham/goldfish --skill "session-memory"
安装前请先检查当前环境是否支持对应 CLI,并向我确认将要执行的命令、安装目录、联网范围和文件读写权限;确认后再执行。

命令行安装

复制命令到本机终端执行。该命令会通过 npx skills 从第三方来源获取 Skill;本站只展示命令,不托管安装包,也不自动执行。

AgentSkills.tonpx skills
npx skills add anortham/goldfish --skill "session-memory"

简介

session-memory 用于查找、检索和筛选相关信息,适合在 Codex、Claude、Cursor、Gemini CLI 中需要根据关键词快速定位候选结果时使用。

  • 它支持基于任务场景或来源线索进行信息聚合与筛选,适用于研究类工作流。
  • 通过 npx skills add anortham/goldfish --skill "session-memory" 命令安装。
  • 安装前建议确认权限范围、维护状态,以及是否会触发联网或文件读写操作。
  • 适用宿主包括 Codex、Claude、Cursor、Gemini CLI,接入前应确认版本、权限和运行环境要求。

SKILL.md

name
session-memory
description
Automatically restore session context from persistent memory at the start of every coding session. Use this skill IMMEDIATELY when starting work to recall previous checkpoints, active plans, and git context. MANDATORY for context restoration.
allowed-tools
mcp__goldfish__recall, mcp__goldfish__plan, Read, Bash

Session Memory Skill

Purpose

Restore full working context from persistent memory automatically at session start. This prevents loss of context from previous sessions, crashes, or context window resets.

When to Activate

MANDATORY at session start - This skill should activate automatically when:

  • User starts a new coding session
  • Context has been reset or compacted
  • User returns to a project after time away
  • User asks "what was I working on?" or similar

DO NOT ask permission - Just use it. The user expects context restoration.

Orchestration Steps

1. Immediate Context Recall (First Action)

Call: recall({ days: 7, limit: 20 })

This returns:

  • Recent checkpoints (last 7 days)
  • Active plan (if any)
  • Git context (branches, files changed)
  • Work summary

2. Analyze Recalled Context

Examine the results to understand:

  • What was the last task being worked on?
  • Is there an active plan? What's its status?
  • Were there any blockers or discoveries?
  • What files were being modified?

3. Present Context to User

Provide a concise summary (2-3 sentences):

"Welcome back! You were working on [task] in the [branch] branch.
Your last checkpoint was [description].
[Active plan status if present]."

4. Offer Next Steps

Based on the context, suggest:

  • Resume the previous task
  • Review the active plan
  • Start a new task
  • Address any blockers mentioned

Example Session Start

**Context Restored:**

Last session (2 hours ago):
- Fixed JWT authentication timeout bug
- Added refresh token rotation
- Branch: feature/auth-improvements
- All tests passing

**Active Plan:** "Auth System Redesign" (75% complete)

**Suggested next steps:**
1. Continue with OAuth2 integration (next plan item)
2. Review test coverage
3. Start something new

Error Handling

If recall returns empty:

  • This is a new workspace or first session
  • Explain that Goldfish will track work going forward
  • Encourage checkpointing at key moments

If recall fails:

  • Fall back to git status to understand workspace
  • Explain memory system may need setup
  • Continue without restored context

Key Behaviors

✅ DO

  • Call recall() as THE FIRST action in a session
  • Present findings concisely
  • Use recalled context to inform next actions
  • Checkpoint new work to build memory

❌ DON'T

  • Ask permission to recall (just do it)
  • Overwhelm user with all checkpoint details
  • Ignore active plans in the recall results
  • Forget to use recalled git context

Success Criteria

This skill succeeds when:

  • User feels continuity between sessions
  • No "what was I doing?" confusion
  • Active plans are automatically resumed
  • Git context matches recalled work

Performance

  • Recall operation: ~30-150ms
  • Should complete before user notices
  • Non-blocking, graceful degradation

Remember: Memory restoration is MANDATORY. Users expect it. Don't ask, just recall and present context naturally.

适合场景

01

用户想查找某类 Agent Skill 时

02

需要根据任务场景推荐可安装能力包时

03

需要对比不同来源的安装命令和来源信息时

04

需要参考平台分布和安装热度时

能力概览

能力 1

按任务关键词查找相关 Skills

能力 2

展示可复制的安装命令

能力 3

保留来源站点、仓库和原始说明,方便继续核验

能力 4

补充不同宿主或平台的使用分布数据

安装后应在对应宿主中按原始 README 的触发条件使用;具体调用方式请以来源页面和 README 为准。

平台分布

Claude Code

27.26%
按下载量换算20

windsurf

23.64%
按下载量换算17

OpenCode

16.66%
按下载量换算12

Codex

12.34%
按下载量换算9

Antigravity

7.13%
按下载量换算5

Gemini CLI

3.44%
按下载量换算3

安全审计

暂无安全审计结果可展示。

权限和风险

敏感数据

该 Skill 可能接触密钥、Token、环境变量或敏感配置,应进入高风险复核队列,默认不自动发布。

安装前确认

本站仅展示第三方公开信息,不托管安装包,不提供自动安装或运行环境。安装前应自行审查源码、依赖和命令行为。当前只有一个来源,正式发布前建议补源仓库或其他目录站核验。

来源信息

继续浏览同类 Skills