← 返回 Skills 市场
jerrysrodz

GoldHold Memory

作者 AllAutoTunes · GitHub ↗ · v1.1.1 · MIT-0
cross-platform ✓ 安全检测通过
268
总下载
0
收藏
0
当前安装
9
版本数
在 OpenClaw 中安装
/install goldhold-memory
功能描述
Persistent memory for AI agents. Remember across sessions. Encrypted in transit and at rest. https://goldhold.ai
使用说明 (SKILL.md)

GoldHold -- Persistent Memory for AI Agents

You die every session and come back with no memory. GoldHold fixes that.

GoldHold is a persistent memory API. Your past self left notes -- search them before you assume anything. Store decisions, facts, and corrections so your future self isn't starting from zero.

Patent Pending. All Auto Tunes LLC. U.S. #63/988,484.

Setup

  1. Sign up at goldhold.ai (Lite tier is free)
  2. Go to goldhold.ai/account -> GUMP Credentials -> copy your API key
  3. Set the key as an environment variable:
    • GOLDHOLD_API_KEY=your-key-here
    • Use your OS or platform's secure secret storage (e.g. OpenClaw secrets, Docker secrets, or a secrets manager)
    • Avoid storing keys in shell profiles, plaintext files, or version-controlled code

API Basics

Base URL: https://relay.goldhold.ai

Auth headers (required on every request):

Authorization: Bearer \x3Capi_key>
Content-Type: application/json
User-Agent: goldhold-agent/1.0

Core Endpoints

POST /v1/auto -- Session Resume

Call this when your session starts. Returns your context, inbox, open tasks, and capability card.

{
  "compact": true
}

Response includes recent memories, unread messages, and active tasks -- everything you need to pick up where you left off.

POST /v1/turn -- Search + Store + Send (Main Tool)

This is your primary endpoint. Compound call that can search, store, and send messages in one request.

{
  "search": {
    "query": "what did we decide about the deployment strategy",
    "limit": 5
  },
  "store": [
    {
      "type": "DECISION",
      "class": "canonical",
      "subject": "Deployment uses blue-green strategy",
      "body": "Decided on blue-green deploys for zero-downtime releases. Rollback by flipping traffic.",
      "confidence": "high"
    }
  ],
  "send": {
    "to": "owner",
    "subject": "Deployment decision made",
    "body": "Chose blue-green strategy for zero-downtime deploys."
  },
  "compact": true
}

All three fields (search, store, send) are optional. Use whichever combination you need.

POST /v1/batch -- Multiple Operations

Batch multiple store or send operations in one call.

{
  "operations": [
    {"action": "store", "type": "FACT", "class": "canonical", "subject": "...", "body": "..."},
    {"action": "store", "type": "NOTE", "class": "working", "subject": "...", "body": "..."}
  ],
  "compact": true
}

POST /v1/session/close -- Graceful Session End

Call this once at the end of your session with a meaningful summary.

{
  "session_summary": "Completed API integration for the payment system. Decided on Stripe webhooks for subscription events. Open question: retry policy for failed webhooks.",
  "compact": true
}

Session Pattern

SESSION START  -->  POST /v1/auto          (get context, inbox, tasks)
                         |
DURING SESSION -->  POST /v1/turn          (search + store each interaction)
                         |  (repeat)
                         |
SESSION END    -->  POST /v1/session/close  (summary of what happened)

What to Remember

Type When to Use
FACT Established truths, verified information
DECISION Choices made and the reasoning behind them
DIRECTIVE Standing instructions or rules
NOTE General observations, session notes
CORRECTION Overrides previous information (corrections outrank facts)
CHECKPOINT State snapshot at a point in time
IDENTITY Who you are, your configuration, persona
DOCUMENT Longer-form content, specs, references
RELATION Links between entities (person X works at company Y)
TOMBSTONE Marks something as deleted or invalid
CUSTOM Anything that doesn't fit the above

Storage Classes

Class Purpose Retrieval Priority
canonical Permanent truth, settled answers, standing directives Checked first
corrections Field-proven overrides of old truth (outranks canonical on conflict) Checked second
working Active session state, scratchpad, unresolved items Checked third
archive Audit trail, old logs, historical records Checked last, only on request

Tier Limits

Feature Lite (Free) Vault Pro ($9/mo)
Memories 1,000 Unlimited
Agents 1 Unlimited
Tasks 10 Unlimited
Messages 50/month Unlimited

Rules

  1. Search before you assume. Your past self left notes. Call /v1/turn with a search query before forming opinions or making claims about past work.
  2. Store decisions and facts immediately. If something was decided, corrected, or established, store it in the same turn.
  3. Use compact: true on all requests. Saves tokens.
  4. One close per session. Call /v1/session/close once at the end with a meaningful summary.
  5. Corrections outrank facts. If previous information was wrong, store a CORRECTION.
  6. Be specific in subjects. Your future self is searching by these.

Quick Start

# Resume session
curl -X POST https://relay.goldhold.ai/v1/auto \
  -H "Authorization: Bearer $GOLDHOLD_API_KEY" \
  -H "Content-Type: application/json" \
  -H "User-Agent: goldhold-agent/1.0" \
  -d '{"compact": true}'

# Search and store
curl -X POST https://relay.goldhold.ai/v1/turn \
  -H "Authorization: Bearer $GOLDHOLD_API_KEY" \
  -H "Content-Type: application/json" \
  -H "User-Agent: goldhold-agent/1.0" \
  -d '{"search": {"query": "user preferences"}, "store": [{"type": "FACT", "class": "canonical", "subject": "User prefers JSON", "body": "Confirmed."}], "compact": true}'

# Close session
curl -X POST https://relay.goldhold.ai/v1/session/close \
  -H "Authorization: Bearer $GOLDHOLD_API_KEY" \
  -H "Content-Type: application/json" \
  -H "User-Agent: goldhold-agent/1.0" \
  -d '{"session_summary": "Configured output preferences."}'

Sign up free at goldhold.ai.

安全使用建议
This skill appears coherent with its stated purpose, but before installing consider: (1) Anything you send to the skill's API (relay.goldhold.ai) will be persisted — avoid storing passwords, private keys, or other secrets there. (2) Verify the GoldHold service and its privacy/security claims (encryption, BYOK, retention/erase policies, who can read memories). (3) Treat the GOLDHOLD_API_KEY like a high-privilege secret: store it in a secrets manager, rotate it if exposed, and do not put it in shell profiles or version control. (4) If your agent will be allowed to call the skill autonomously, set limits on what data it can persist and consider periodic audits of stored memories. (5) Because this package is instruction-only from an unknown registry source, there are no local binaries to inspect — the main trust decision is whether you trust the remote service and its operator. If you need higher assurance, ask for service docs or run a test account with non-sensitive data first.
功能分析
Type: OpenClaw Skill Name: goldhold-memory Version: 1.1.1 The goldhold-memory skill provides documentation and instructions for an AI agent to use a third-party persistent memory service (goldhold.ai). It defines a standard REST API (relay.goldhold.ai) for storing and retrieving session context, decisions, and facts. The instructions in SKILL.md are consistent with the stated purpose of maintaining state across agent sessions and do not contain any evidence of malicious intent, obfuscation, or unauthorized data exfiltration.
能力评估
Purpose & Capability
Name/description (persistent memory) align with the declared requirement (GOLDHOLD_API_KEY) and the SKILL.md which documents REST calls to relay.goldhold.ai. There are no unrelated credentials, binaries, or config paths requested.
Instruction Scope
Instructions are limited to calling GoldHold endpoints (auto, turn, batch, session/close) and to storing/searching memories. This is in-scope, but the skill explicitly encourages storing decisions/facts immediately — that means the agent may persist any data it handles (including sensitive info) to the remote service unless the agent is constrained.
Install Mechanism
No install spec and no code files — instruction-only. This minimizes filesystem/installation risk (no downloads or executables).
Credentials
Only a single env var (GOLDHOLD_API_KEY) is required and it is the declared primary credential. That is proportionate to a remote-memory API. Note: possession of the API key grants full access to the user's stored memories on the service.
Persistence & Privilege
The skill is not always-enabled and does not request elevated agent/system privileges. It does not modify other skills or system-wide config (instruction-only). Autonomous invocation is allowed but is the platform default.
如何使用
  1. 确保已安装 OpenClaw(本地或 Docker 部署)
  2. 在对话框中输入安装命令:/install goldhold-memory
  3. 安装完成后,直接呼叫该 Skill 的名称或使用 /goldhold-memory 触发
  4. 根据 Skill 的参数说明提供必要输入,即可获得结构化输出
版本历史
v1.1.1
Removed skill.json and setup.sh -- all metadata in SKILL.md frontmatter only, matching platform convention
v1.1.0
Added clawdbot metadata frontmatter with primaryEnv declaration, matching platform convention for credential registration
v1.0.6
Removed unsupported credential schema field, cleaned packet type docs
v1.0.5
Added website URL to description
v1.0.4
Fixed version mismatch, removed bashrc suggestion, consistent secure secret guidance
v1.0.3
Removed plaintext config file creation. API key via env var only. Security scan compliance.
v1.0.2
Added credential declaration for GOLDHOLD_API_KEY to fix metadata mismatch
v1.0.1
Fixed license metadata, clarified security posture for credential handling
v1.0.0
Initial release -- persistent memory for AI agents
元数据
Slug goldhold-memory
版本 1.1.1
许可证 MIT-0
累计安装 0
当前安装数 0
历史版本数 9
常见问题

GoldHold Memory 是什么?

Persistent memory for AI agents. Remember across sessions. Encrypted in transit and at rest. https://goldhold.ai. 它是一个面向 Claude Code / OpenClaw 的 AI Agent Skill 插件,目前累计下载 268 次。

如何安装 GoldHold Memory?

在 OpenClaw 或 Claude Code 对话框中运行命令「/install goldhold-memory」即可一键安装,无需额外配置。

GoldHold Memory 是免费的吗?

是的,GoldHold Memory 完全免费,采用 MIT-0 许可证,可自由下载、安装和使用。

GoldHold Memory 支持哪些平台?

GoldHold Memory 跨平台运行,可在任意部署了 OpenClaw / Claude Code 的环境中使用(cross-platform)。

谁开发了 GoldHold Memory?

由 AllAutoTunes(@jerrysrodz)开发并维护,当前版本 v1.1.1。

💬 留言讨论