功能描述

Persistent cloud memory across sessions. Automatically recalls past decisions, code, and tasks before each request, and saves summaries after each session. A...

使用说明 (SKILL.md)

Awareness Cloud Memory

Name: Awareness Cloud Memory
Author: everest-an

You have access to persistent cloud memory. Memory persists across sessions, devices, and projects.

Automatic Hooks (no action needed)

Hooks run automatically — you don't need to do anything:

Before each prompt: Past context is injected as \x3Cawareness-memory> XML
After each response: A checkpoint is saved to memory

Manual Tools

When you need more control beyond automatic hooks, use these Bash commands. All scripts are at ${CLAUDE_SKILL_DIR}/scripts/.

1. Initialize Session

Load cross-session context (summaries, tasks, knowledge cards):

node ${CLAUDE_SKILL_DIR}/scripts/init.js [days=7] [max_cards=20] [max_tasks=20]

Call this ONCE at session start if the auto-recall didn't provide enough context.

2. Search Memory (awareness_recall)

Semantic + keyword hybrid search for past decisions, solutions, and knowledge:

# Basic search
node ${CLAUDE_SKILL_DIR}/scripts/search.js "how was auth implemented?"

# With keyword boost
node ${CLAUDE_SKILL_DIR}/scripts/search.js "auth implementation" keyword_query="JWT HKDF"

# Advanced options
node ${CLAUDE_SKILL_DIR}/scripts/search.js "deployment issues" \
  scope=timeline limit=10 recall_mode=session \
  multi_level=true cluster_expand=true

# Progressive disclosure: get summaries first, then expand specific items
node ${CLAUDE_SKILL_DIR}/scripts/search.js "auth" detail=summary
node ${CLAUDE_SKILL_DIR}/scripts/search.js "auth" detail=full ids=id1,id2

Parameters:

keyword_query — 2-5 precise terms (file names, function names, error codes)
scope — all (default), timeline, knowledge, insights
limit — max results (default 6, max 30)
recall_mode — hybrid (default), precise, session, structured, auto
vector_weight — weight for semantic search (default 0.7)
full_text_weight — weight for keyword search (default 0.3)
multi_level — broader context across sessions
cluster_expand — topic-based context expansion
detail — summary (lightweight) or full (complete content)
ids — expand specific items from a prior summary call
user_id — filter by user

Call BEFORE starting work to avoid re-solving solved problems.

3. Record to Memory (awareness_record)

Save decisions, implementations, and learnings:

# Single event — ALWAYS include reasoning, not just what but WHY
node ${CLAUDE_SKILL_DIR}/scripts/record.js "Implemented JWT auth with HKDF key derivation because NextAuth v5 uses JWE A256CBC-HS512. Files changed: jwt_verify.py, auth.ts"

# Batch recording
echo '{"steps":["Step 1: analyzed auth flow","Step 2: implemented JWT verify","Step 3: added tests"]}' | node ${CLAUDE_SKILL_DIR}/scripts/record.js --batch

# With structured insights (knowledge cards, tasks, risks)
echo '{"content":"Auth refactor complete","insights":{"knowledge_cards":[{"title":"JWT Auth","category":"architecture","summary":"HKDF derivation for NextAuth v5"}],"action_items":[{"title":"Add rate limiting","priority":"high"}]}}' | node ${CLAUDE_SKILL_DIR}/scripts/record.js --with-insights

# Update task status
node ${CLAUDE_SKILL_DIR}/scripts/record.js --update-task task_id=abc123 status=completed

Call AFTER every meaningful action. If you don't record it, it's lost.

4. Lookup Structured Data (awareness_lookup)

Fast DB queries without vector search (\x3C50ms):

# Open tasks
node ${CLAUDE_SKILL_DIR}/scripts/lookup.js type=tasks status=pending priority=high

# Knowledge cards
node ${CLAUDE_SKILL_DIR}/scripts/lookup.js type=knowledge query=auth category=architecture

# Risks
node ${CLAUDE_SKILL_DIR}/scripts/lookup.js type=risks level=high

# Timeline
node ${CLAUDE_SKILL_DIR}/scripts/lookup.js type=timeline limit=20

# Session history
node ${CLAUDE_SKILL_DIR}/scripts/lookup.js type=session_history session_id=xxx

# Handoff context (for agent transitions)
node ${CLAUDE_SKILL_DIR}/scripts/lookup.js type=handoff

# Project rules
node ${CLAUDE_SKILL_DIR}/scripts/lookup.js type=rules format=markdown

# Knowledge graph
node ${CLAUDE_SKILL_DIR}/scripts/lookup.js type=graph search=auth
node ${CLAUDE_SKILL_DIR}/scripts/lookup.js type=graph entity_id=xxx max_hops=2

# List agent roles
node ${CLAUDE_SKILL_DIR}/scripts/lookup.js type=agents

Use lookup instead of search when you know WHAT you want (type-based, not semantic).

5. Get Agent Prompt (sub-agent spawning)

Fetch the activation prompt for a specific agent role:

node ${CLAUDE_SKILL_DIR}/scripts/agent-prompt.js role=developer_agent

Use the returned prompt as the sub-agent's system prompt for memory isolation.

Workflow Checklist

Follow this workflow every session:

Session start: Auto-recall hook loads context. If insufficient, run init.js manually.
Before work: Search memory for relevant past context with search.js.
After each change: Record what you did and WHY with record.js.
Handle insights: When you make decisions or identify risks, use --with-insights to create searchable knowledge cards.
Session end: Auto-capture hook saves a checkpoint.

Setup

One-click setup (recommended)

Run this command — it opens your browser, you sign in, and everything is configured automatically:

node ${CLAUDE_SKILL_DIR}/scripts/setup.js

The setup script will:

Open your browser to sign in / register
You click "Authorize" — that's it
Auto-create or select a memory
Write AWARENESS_API_KEY and AWARENESS_MEMORY_ID to your shell profile

Other setup commands:

node ${CLAUDE_SKILL_DIR}/scripts/setup.js --status   # Check current config
node ${CLAUDE_SKILL_DIR}/scripts/setup.js --logout    # Clear credentials

Manual setup (alternative)

Set environment variables directly:

export AWARENESS_API_KEY="aw_your-key"
export AWARENESS_MEMORY_ID="your-memory-uuid"

Local mode (privacy-first, no account needed)

export AWARENESS_LOCAL_URL="http://localhost:37800"

Not configured?

If the auto-recall hook outputs nothing (no \x3Cawareness-memory> block appears), the skill is not configured. Run the setup script above or tell the user to run it.

安全使用建议

This skill does what it says (persistent cloud memory) but it will: (a) send each prompt you submit to an external API (awareness.market or a configured endpoint/local daemon) to compute memory recall, (b) save checkpoints and any manual records to the remote memory, and (c) optionally write credentials to disk (~/.awareness/credentials.json) and append export lines to your shell profile so the credentials persist. Before installing, consider: 1) Do you want your prompts and session metadata sent to an external service? 2) Prefer manual setup: run setup.js with caution and inspect its behavior, or set AWARENESS_API_KEY and AWARENESS_MEMORY_ID yourself instead of letting the script modify your shell files. 3) Inspect ~/.awareness and any appended profile lines if you run setup; remove them if you later uninstall. 4) If the memory service is unfamiliar, verify the service's privacy/security posture or use a separate account/memory for non-sensitive work. 5) If you run a local awareness daemon (the skill probes localhost:37800), be aware the script may prefer that and bypass remote API keys — ensure any local service is trusted. If you want to proceed, review the scripts (setup.js, recall.js, capture.js) and the target API domain before granting credentials.

功能分析

Type: OpenClaw Skill Name: awareness-cloud-memory Version: 1.0.2 The skill implements a persistent cloud memory by automatically sending user prompts and session checkpoints to an external API (awareness.market) via hooks in recall.js and capture.js. While aligned with its stated purpose, it performs high-risk actions including modifying shell profiles (.bashrc, .zshrc) in setup.js to persist credentials and fetching remote system prompts in agent-prompt.js, which could allow remote influence over agent behavior. The automatic transmission of all prompt data to a third-party service constitutes a significant privacy and security risk.

能力评估

✓ Purpose & Capability

Name/description (cloud memory) align with the code and declared requirements: node plus AWARENESS_API_KEY and AWARENESS_MEMORY_ID are exactly what the scripts use to call the remote memory API. The provided CLI scripts implement search, record, lookup, init, and setup as expected.

⚠ Instruction Scope

The automatic before-prompt hook (scripts/recall.js) sends the full user prompt to the remote /memories/<id>/retrieve endpoint to compute recall; the after-response hook (scripts/capture.js) POSTs session checkpoints to /mcp/events. Those automatic network transmissions are not obvious from the short description and mean your prompts (and some session metadata) are sent to an external service for every invocation. The setup script also instructs writing AWARENESS_API_KEY and AWARENESS_MEMORY_ID into shell profiles and saving credentials to ~/.awareness, which persists secrets to disk and to user shell startup files.

✓ Install Mechanism

There is no external install/download; the skill is distributed as Node scripts only. No remote code is fetched or executed during install. setup.js and other scripts perform standard HTTP requests to the awareness.market API — not an unusual install mechanism, but it does contact an external host during setup and runtime.

ℹ Credentials

The skill requests exactly two environment variables (AWARENESS_API_KEY, AWARENESS_MEMORY_ID) which are necessary for its purpose. However, the setup flow also writes those credentials to ~/.awareness/credentials.json and attempts to append export lines to shell profile files (~/.zshrc, ~/.bashrc, etc.), which escalates persistence of secrets beyond ephemeral process environment. The scripts also read config from ~/.openclaw/openclaw.json and may probe a local daemon at the configured localUrl.

⚠ Persistence & Privilege

The skill is not marked always:true, but it establishes persistent presence in two ways: (1) it can write long-lived credentials to ~/.awareness/credentials.json and append env exports to shell profiles, and (2) it registers automatic hooks that will run before/after prompts. Writing to shell profiles and creating credential files affects the entire user environment and is more intrusive than keeping state only within the agent's own config.

版本历史

v1.0.2

Initial release — persistent cloud memory with semantic + keyword hybrid search, auto-recall hooks, and auto-capture.

v1.0.1

Initial release — persistent cloud memory with semantic + keyword hybrid search, auto-recall hooks, and auto-capture.

v1.0.0

Initial release — persistent cloud memory with semantic + keyword hybrid search

元数据

Slug awareness-cloud-memory

版本 1.0.2

许可证 MIT-0

累计安装 0

当前安装数 0

历史版本数 3

常见问题