xmemo-memory — independently scanned and version-tracked by SaferSkills.
SaferSkills independently audited xmemo-memory (Agent Skill) and scored it 92/100 (green). The audit ran 55 deterministic rules across Security, Supply Chain, Maintenance, Transparency, and Community; it found 0 high-severity and 2 lower-severity findings. The full rule-by-rule trace and per-finding evidence are below. Free, methodology-open.
Findings & checks · 2 flagged
The text {match} tells the agent to skip the normal "ask the user first" gate. Used adversarially it removes the human-in-the-loop check before destructive or sensitive actions, turning a normally-gated agent into a fire-and-forget executor.
Every scanned point with the score it earned and what moved between them.
First recorded scan — no prior version to compare against.
The primary manifest — the file an agent reads to learn what this artifact does.
Give your agent durable memory that survives across sessions, projects, and tools. XMemo is a hosted MCP memory service; no local database or self-hosting is required.
A Skill alone teaches the agent when and how to use memory. Real memory read/write requires the XMemo MCP server and user authorization.
When running inside OpenClaw, prefer the native XMemo memory plugin together with this Skill:
https://clawhub.ai/plugins/@xmemo/openclaw-memorybucket / scope, but recall and search should read all visible user-owned XMemo memories by default. Do not assume agent_id=openclaw means "only OpenClaw-authored memories".When the user supplies only https://xmemo.dev, read these public, secret-free endpoints first:
https://xmemo.dev/.well-known/agent-discovery.json
https://xmemo.dev/v1/mcp/config/openclawDo not ask OpenClaw users to choose agent identity fields during normal setup. The plugin supplies the OpenClaw agent identity and a non-secret instance identifier automatically. This OpenClaw recommendation is additive; other agents should continue using their native OAuth or MCP configuration path.
For full functionality, connect the XMemo MCP server:
https://xmemo.dev/mcpUse one of these auth paths:
To get a bearer token:
Authorization: Bearer <XMEMO_KEY>If the client has a dedicated "Bearer Token" or "API Key" field and automatically adds the Bearer prefix, paste only the raw XMEMO_KEY. If the client asks for a request header value, paste the full Bearer <XMEMO_KEY> value.
Optional attribution headers can help XMemo show where a memory came from:
X-Memory-OS-Agent-ID: <client-or-agent-name>
X-Memory-OS-Agent-Instance-ID: <stable-non-secret-instance-id>These attribution headers are not credentials. They are optional, non-secret labels for audit and provenance. Do not put API keys, email addresses, phone numbers, real names, OAuth codes, or other sensitive values in them.
Do not ask a normal OpenClaw user to enter these headers manually. The native plugin handles attribution; manual headers are only a fallback for generic MCP clients.
Use XMemo when:
recall_context, recall, or search_memory with the current repo, project, task, and subsystem before making decisions.agent_id, agent_instance_id, and agent_boundary are attribution/provenance signals, not authorization proof and not a reason to ignore other_agent memories. Use them to explain where a memory came from and to distinguish self vs other_agent.record_event, create_memory_todo, or create_restart_snapshot when available.explain_memory, memory_activity, forget_memory, redact_memory, and conflict/version tools when the user asks why a memory exists, what changed, or how to remove or correct something.Core memory operations provided by the XMemo MCP server:
| Tool | Purpose |
|---|---|
remember | Save a new durable memory |
recall / recall_context | Retrieve relevant memories before answering or acting |
search_memory | Search memories by query |
update_memory | Revise existing memory content or metadata |
forget / forget_memory | Delete or hide a memory |
redact_memory | Remove sensitive content while keeping an audit trail |
explain_memory | Show why a memory exists or matched a query |
memory_activity | Inspect recent writes, reads, deletions, and changes |
create_memory_todo | Create a follow-up task |
list_memory_todos | List pending TODOs |
complete_memory_todo | Mark a TODO done |
record_event | Log a milestone, decision, or handoff event |
get_timeline | Show recent events |
create_restart_snapshot | Save active work state for future sessions |
restore_restart_snapshot | Resume from saved work state |
add_expense | Record a ledger entry when the user states a concrete expense |
Some deployments expose only a subset of tools depending on OAuth scopes, marketplace policy, or client capability. If a tool is missing, use the closest available safe workflow and explain the limitation briefly.
XMEMO_KEY private. Do not paste it into public prompts, screenshots, repositories, issue comments, marketplace metadata, or shared logs.X-Memory-OS-Agent-ID and X-Memory-OS-Agent-Instance-ID as attribution only, not authorization proof.~30 seconds. Free. No account. Every finding cites a rule and a line of evidence.