aionis_memory_policy_loop

# Aionis Memory Policy Loop Skill

Use this skill when the user asks for long-term memory, retrieval with citations, tool routing from memory rules, or feedback-driven policy adaptation.

## Requirements (Local Standalone)

Environment variables:

1. `AIONIS_BASE_URL`:
   - host run: `http://127.0.0.1:3001`
   - container-to-host run: `http://host.docker.internal:3001`
2. One auth method:
   - `AIONIS_API_KEY`
   - or `AIONIS_AUTH_BEARER`
3. Optional:
   - `AIONIS_TENANT_ID` (default: `default`)
   - `AIONIS_SCOPE_PREFIX` (default: `clawbot`)

## Safety Rules

1. Never print full secrets in responses.
2. Keep scope fixed per project: `clawbot:<project>`.
3. Do not write raw tool output dumps into memory; store concise summaries.
4. Keep requests bounded: set limits for recall and context assembly.
5. If `/v1/memory/context/assemble` is unavailable, fallback to `/v1/memory/recall_text` and continue.

## Connectivity Precheck

Before running the memory loop, ensure Aionis standalone is reachable:

1. `GET /health` returns `200`.
2. `POST /v1/memory/write` with `x-api-key` returns `200`.
3. If check fails, stop and return a clear connectivity/auth error.

## Auto Bootstrap Command

If local standalone is not running, execute:

```bash
bash ./bootstrap-local-standalone.sh
```

Then load runtime env:

```bash
source ./.runtime/clawbot.env
```

## Default Workflow

1. Ingest key facts/results:
   - `POST /v1/memory/write`
2. Build layered context before planning:
   - `POST /v1/memory/context/assemble`
   - fallback to `POST /v1/memory/recall_text` if assemble endpoint is unavailable
3. Route tools with policy:
   - `POST /v1/memory/tools/select`
4. Close the loop after execution:
   - `POST /v1/memory/tools/feedback`

## Request Templates

Use these templates (replace placeholders):

### write

```json
{
  "tenant_id": "default",
  "scope": "clawbot:demo-project",
  "input_text": "Customer prefers email follow-up",
  "auto_embed": true,
  "nodes": [
    {
      "client_i...

# ClawHub Publish Notes

This folder is a publish-ready OpenClaw skill package.

## Contents

1. `SKILL.md` - skill definition and runtime guidance.
2. `AIONIS_ENV_TEMPLATE.txt` - local standalone defaults.
3. `bootstrap-local-standalone.sh` - one-click local install/start script.

## Local Standalone First (Required)

One-click bootstrap:

```bash
bash ./bootstrap-local-standalone.sh
```

The script creates:

1. `.runtime/aionis.env` for container runtime
2. `.runtime/clawbot.env` for Clawbot skill variables

Manual start (equivalent):

```bash
docker run -d --name aionis-standalone-local --restart unless-stopped \
  -p 127.0.0.1:3001:3001 \
  --env-file ./.runtime/aionis.env \
  -v aionis-standalone-data:/var/lib/postgresql/data \
  ghcr.io/cognary/aionis:standalone-v0.2.5
```

Health check:

```bash
curl -fsS http://127.0.0.1:3001/health
```

Auth check (`x-api-key`):

```bash
curl -sS http://127.0.0.1:3001/v1/memory/write \
  -H 'content-type: application/json' \
  -H 'x-api-key: YOUR_MEMORY_API_KEY' \
  -d '{"tenant_id":"default","scope":"default","input_text":"clawbot local standalone test"}'
```

## Publish

```bash
npm i -g clawhub
clawhub login
clawhub whoami

clawhub publish . \
  --slug aionis-memory-policy-loop \
  --name "Aionis Memory Policy Loop" \
  --version 1.0.0 \
  --changelog "Initial release" \
  --tags latest
```

## Update

```bash
clawhub publish . \
  --slug aionis-memory-policy-loop \
  --name "Aionis Memory Policy Loop" \
  --version 1.0.1 \
  --changelog "Fix prompt and examples" \
  --tags latest
```