Hindclaw

Name: Hindclaw
Rating: 3.5 (1 reviews)
Author: mrkhachaturov

By mrkhachaturov ⭐ 5 stars 👁 763 views ▲ 0 votes

Production-grade Hindsight memory plugin for OpenClaw — per-agent config, multi-bank recall, IaC bank management

GitHub

Install

openclaw plugins install hindclaw

README

<p align="center">
  <img src=".github/assets/hindclaw.png" alt="HindClaw">
</p>

<p align="center">
  Self-hosted Hindsight management platform — multi-tenant access control, user/group permissions, client integrations, and infrastructure tooling for AI agent memory.
</p>

<p align="center">
  <a href="https://www.npmjs.com/package/hindclaw"><img src="https://img.shields.io/npm/v/hindclaw?style=flat-square&color=0f766e" alt="npm"></a>
  <a href="https://pypi.org/project/hindclaw-extension/"><img src="https://img.shields.io/pypi/v/hindclaw-extension?style=flat-square&color=0f766e" alt="PyPI"></a>
  <img src="https://img.shields.io/badge/license-MIT-10b981?style=flat-square" alt="License">
</p>

<p align="center">
  <a href="https://hindclaw.pro">Documentation</a> &middot;
  <a href="https://hindsight.vectorize.io">Hindsight</a> &middot;
  <a href="https://openclaw.ai">OpenClaw</a>
</p>

---

## Why HindClaw?

Built on [Hindsight](https://hindsight.vectorize.io) — the highest-scoring agent memory system on the [LongMemEval benchmark](https://vectorize.io/#:~:text=The%20New%20Leader%20in%20Agent%20Memory).

The official Hindsight plugin gives you auto-capture and auto-recall. HindClaw adds what you need to run it in production — two orthogonal dimensions that combine per-message:

**WHO** (permissions) — resolved through 4 layers, each can override any field:
- Global config defaults → group merge → bank group override → bank user override
- Not just `recall: true/false` — LLM model, token budget, extraction depth, tag visibility, retention frequency. 11 configurable fields at every layer.

**HOW** (strategies) — resolved per topic:
- Each conversation topic routes to a named strategy with its own extraction mission, mode, and entity labels
- Strategies are orthogonal to permissions — a blocked user never reaches strategy resolution

```mermaid
graph TD
    MSG["💬 Message: user + bank + topic"] --> PERM["🔐 Resolve permissions<br/>for THIS user on THIS bank"]
    MSG --> STRAT["🎯 Resolve strategy<br/>for THIS topic on THIS bank"]
    PERM --> R{"recall?"}
    PERM --> W{"retain?"}
    R -->|true| RECALL["📥 Recall<br/>budget, tokens, tag filters<br/>from resolved permissions"]
    R -->|false| NO_R["🚫 No recall"]
    W -->|true| RETAIN["📤 Retain<br/>roles, tags, LLM model<br/>+ topic strategy"]
    W -->|false| NO_W["🚫 No retain"]
    STRAT --> RETAIN

    style MSG fill:#1d4ed8,color:#fff,stroke:#1d4ed8
    style PERM fill:#8b5cf6,color:#fff,stroke:#8b5cf6
    style STRAT fill:#c2410c,color:#fff,stroke:#c2410c
    style RECALL fill:#10b981,color:#fff,stroke:#10b981
    style RETAIN fill:#10b981,color:#fff,stroke:#10b981
    style NO_R fill:#ef4444,color:#fff,stroke:#ef4444
    style NO_W fill:#f59e0b,color:#fff,stroke:#f59e0b
```

Every combination of **(user × bank × topic)** can produce different behavior — different access flags, different LLM model, different recall budget, different extraction strategy.

---

### 🔐 Per-User Access & Behavioral Overrides

The same user gets different behavior on different agents. Every parameter is configurable per group, overridable per bank and per user.

```mermaid
graph TD
    MSG["💬 Message"] --> ID["🔍 Resolve identity"]
    ID --> GROUPS["👥 Merge group permissions"]
    GROUPS --> BANK{"🏦 Bank overrides?"}
    BANK -->|group override| BG["⚙️ Bank group permissions"]
    BANK -->|user override| BU["👤 Bank user permissions"]
    BANK -->|none| GLOBAL["📋 Global defaults"]
    BG --> RESOLVED["✅ Resolved permissions"]
    BU --> RESOLVED
    GLOBAL --> RESOLVED
    RESOLVED --> TOPIC{"🎯 Topic?"}
    TOPIC -->|mapped| STRATEGY["📝 Named strategy"]
    TOPIC -->|unmapped| DEFAULT["📝 Bank defaults"]

    style MSG fill:#1d4ed8,color:#fff,stroke:#1d4ed8
    style ID fill:#8b5cf6,color:#fff,stroke:#8b5cf6
    style GROUPS fill:#8b5cf6,color:#fff,stroke:#8b5cf6
    style BG fill:#c2410c,color:#fff,stroke:#c2410c
    style BU fill:#c2410c,color:#fff,stroke:#c2410c
    style GLOBAL fill:#c2410c,color:#fff,stroke:#c2410c
    style RESOLVED fill:#0f766e,color:#fff,stroke:#0f766e
    style STRATEGY fill:#10b981,color:#fff,stroke:#10b981
    style DEFAULT fill:#f59e0b,color:#fff,stroke:#f59e0b
```

**Override chain** (most specific wins):

```
config.json5 defaults → group → bank group → bank user
```

**Configurable at every level:** `recall`, `retain`, `retainRoles`, `retainTags`, `retainEveryNTurns`, `recallBudget`, `recallMaxTokens`, `recallTagGroups`, `llmModel`, `llmProvider`, `excludeProviders`

**Same user, different agents:**

| | agent-1 (strategic) | agent-2 (financial) |
|---|---|---|
| **user-1** (executive) | recall + retain, high budget, no filters | recall + retain, high budget, no filters |
| **user-2** (dept-head) | recall only, mid budget, filtered | recall + retain, high budget (user override) |
| **user-3** (staff) | blocked (no entry → `_default`) | recall only, low budget, filtered |
| **anonymous** | blocked | blocked |

### 🔀 Cross-Agent Recall

One agent queries multiple banks in parallel. Permissions checked per-bank.

```mermaid
graph LR
    Q["🔍 agent-1 recall query"] --> B1["🏦 bank: agent-1"]
    Q --> B2["🏦 bank: agent-2"]
    Q --> B3["🏦 bank: agent-3"]
    B1 -->|recall: true| R1["📥 results"]
    B2 -->|recall: true| R2["📥 results"]
    B3 -->|recall: false| SKIP["🚫 skipped"]
    R1 --> MERGE["🔀 Merge + interleave"]
    R2 --> MERGE
    MERGE --> INJECT["💉 Inject into prompt"]

    style Q fill:#1d4ed8,color:#fff,stroke:#1d4ed8
    style B1 fill:#8b5cf6,color:#fff,stroke:#8b5cf6
    style B2 fill:#8b5cf6,color:#fff,stroke:#8b5cf6
    style B3 fill:#8b5cf6,color:#fff,stroke:#8b5cf6
    style R1 fill:#10b981,color:#fff,stroke:#10b981
    style R2 fill:#10b981,color:#fff,stroke:#10b981
    style SKIP fill:#ef4444,color:#fff,stroke:#ef4444
    style MERGE fill:#0f766e,color:#fff,stroke:#0f766e
    style INJECT fill:#0f766e,color:#fff,stroke:#0f766e
```

### 🎯 Named Retain Strategies

Different conversation topics routed to different extraction strategies.

```mermaid
graph LR
    MSG["💬 Incoming message"] --> TOPIC{"🎯 Topic ID?"}
    TOPIC -->|280304| DEEP["🔬 deep-analysis"]
    TOPIC -->|280418| LIGHT["⚡ lightweight"]
    TOPIC -->|other| DEFAULT["📋 bank default"]
    DEEP --> RETAIN1["📝 Verbose extraction"]
    LIGHT --> RETAIN2["📝 Concise extraction"]
    DEFAULT --> RETAIN3["📝 Standard extraction"]

    style MSG fill:#1d4ed8,color:#fff,stroke:#1d4ed8
    style DEEP fill:#8b5cf6,color:#fff,stroke:#8b5cf6
    style LIGHT fill:#f59e0b,color:#fff,stroke:#f59e0b
    style DEFAULT fill:#c2410c,color:#fff,stroke:#c2410c
    style RETAIN1 fill:#10b981,color:#fff,stroke:#10b981
    style RETAIN2 fill:#10b981,color:#fff,stroke:#10b981
    style RETAIN3 fill:#10b981,color:#fff,stroke:#10b981
```

### 🏗️ Infrastructure as Code

`hindclaw plan` shows what will change. `hindclaw apply` syncs it. Like Terraform for memory banks.

```mermaid
graph LR
    FILE["📂 Bank config files"] --> PLAN["📋 hindclaw plan"]
    PLAN --> DIFF{"🔍 Changes?"}
    DIFF -->|none| OK["✅ Up to date"]
    DIFF -->|yes| SHOW["📄 Show diff"]
    SHOW --> APPLY["⚡ hindclaw apply"]
    APPLY --> CONFIRM{"❓ Confirm?"}
    CONFIRM -->|yes| SYNC["🚀 Sync to Hindsight"]
    CONFIRM -->|no| CANCEL["❌ Cancelled"]

    style FILE fill:#1d4ed8,color:#fff,stroke:#1d4ed8
    style PLAN fill:#8b5cf6,color:#fff,stroke:#8b5cf6
    style APPLY fill:#8b5cf6,color:#fff,stroke:#8b5cf6
    style OK fill:#10b981,color:#fff,stroke:#10b981
    style SHOW fill:#f59e0b,color:#fff,stroke:#f59e0b
    style SYNC fill:#10b981,color:#fff,stroke:#10b981
    style CANCEL fill:#ef4444,color:#fff,stroke:#ef4444
```

### 🧩 Session Start Context

Mental models loaded before the first message — no cold start.

```mermaid
graph LR
    START["🎬 Session starts"] --> LOAD["📦 Load mental models"]
    LOAD --> M1["🧠 Project context"]
    LOAD --> M2["👤 User preferences"]
    M1 --> INJECT["💉 Inject into system prompt"]
    M2 --> INJECT
    INJECT --> READY["✅ Agent ready with full context"]
    READY --> MSG1["💬 First user message"]

    style START fill:#1d4ed8,color:#fff,stroke:#1d4ed8
    style LOAD fill:#8b5cf6,color:#fff,stroke:#8b5cf6
    style M1 fill:#c2410c,color:#fff,stroke:#c2410c
    style M2 fill:#c2410c,color:#fff,stroke:#c2410c
    style INJECT fill:#0f766e,color:#fff,stroke:#0f766e
    style READY fill:#10b981,color:#fff,stroke:#10b981
    style MSG1 fill:#10b981,color:#fff,stroke:#10b981
```

### 🪞 Reflect on Recall

Instead of raw memory retrieval, the agent reasons over its memories.

```mermaid
graph LR
    Q["💬 User question"] --> MODE{"🪞 Reflect enabled?"}
    MODE -->|yes| REFLECT["🧠 Hindsight reflect API"]
    MODE -->|no| RECALL["📥 Hindsight recall API"]
    REFLECT --> REASON["💡 LLM reasons over memories"]
    REASON --> ANSWER["✅ Grounded response"]
    RECALL --> RAW["📋 Raw memory list"]
    RAW --> ANSWER

    style Q fill:#1d4ed8,color:#fff,stroke:#1d4ed8
    style REFLECT fill:#8b5cf6,color:#fff,stroke:#8b5cf6
    style RECALL fill:#c2410c,color:#fff,stroke:#c2410c
    style REASON fill:#0f766e,color:#fff,stroke:#0f766e
    style RAW fill:#f59e0b,color:#fff,stroke:#f59e0b
    style ANSWER fill:#10b981,color:#fff,stroke:#10b981
```

### 🌐 Multi-Server Support

Per-agent infrastructure routing — one gateway, multiple Hindsight servers.

```mermaid
graph LR
    GW["🌐 Gateway"] --> A1["🤖 agent-1"]
    GW --> A2["🤖 agent-2"]
    GW --> A3["🤖 agent-3"]
    GW --> A4["🤖 agent-4"]
    A1 --> HOME["🏠 Home server"]
    A2 --> HOME
    A3 --> OFFICE["🏢 Office server"]
    A4 --> LOCAL["💻 Local daemon"]

    style GW fill:#0f766e,color:#fff,stroke:#0f766e
    style A1 fill:#1d4ed8,color:#fff,stroke:#1d4ed8
    style A2 fill:#1d4ed8,color:#fff,stroke:#1d4ed8
    style A3 fill:#1d4ed8,color:#fff,stroke:#1d4ed8
    style A4 fill:#1d4ed8,color:#fff,stroke:#1d4ed8
    style HOME fill:#8b5cf6,color:#fff,stroke:#8b5cf6
    style OFFICE fill:#c2410c,color:#fff,st

... (truncated)

voice