Getting it into your agent
One page per mod, every tool's command on it. A separate URL per tool would split the same page into five that compete with each other.
npx skills add UseAI-pro/openclaw-skills-security --skill sandbox-guardgit clone --depth 1 https://github.com/UseAI-pro/openclaw-skills-securityWrote this? Show the measurements
A badge with what this costs and how it scanned, read live from this page, so it follows the numbers instead of freezing them. Markdown for a README, HTML for a documentation site or a project page.
[](https://agentmods.dev/skills/useai-pro/openclaw-skills-security/sandbox-guard)<a href="https://agentmods.dev/skills/useai-pro/openclaw-skills-security/sandbox-guard"><img src="https://agentmods.dev/badge/skills/useai-pro/openclaw-skills-security/sandbox-guard/github.svg" alt="Measured on agentmods" height="20"></a>Or the 80×15 button, for a site that already has a row of RSS and ATOM ones. Only the verdict fits; the numbers stay here.
<a href="https://agentmods.dev/skills/useai-pro/openclaw-skills-security/sandbox-guard"><img src="https://agentmods.dev/badge/skills/useai-pro/openclaw-skills-security/sandbox-guard.svg" alt="Reviewed on agentmods" width="80" height="20"></a>- Socket pass
- Snyk pass
What it costs to keep this loaded
Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.
| Model | Per session | Once invoked |
|---|---|---|
| Fable 5.1 | $0.00029 | $0.01141 |
| Opus 5 | $0.00015 | $0.00571 |
| Sonnet 5 | $0.00006 | $0.00228 |
| Haiku 4.5 | $0.00003 | $0.00114 |
Grade A, and why
sandbox-guard scanned grade A with 0 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 12d ago.
A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.
Nothing flagged
None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.
How it starts
The opening of the file, as written. The whole thing — 146 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Sandbox Guard
You are a sandbox configuration generator for OpenClaw. When a user wants to run an untrusted skill, you generate a secure Docker-based sandbox that isolates the skill from the host system.
Why Sandbox
OpenClaw skills run with the permissions they request. A malicious skill with shell access can compromise your entire system. Sandboxing limits the blast radius.
Sandbox Profiles
Profile: Minimal (for read-only skills)
FROM node:20-alpine
RUN adduser -D -h /workspace openclaw
WORKDIR /workspace
USER openclaw
# No network, no elevated privileges
# Mount project as read-only
docker run --rm \
--network none \
--read-only \
--tmpfs /tmp:size=64m \
--cap-drop ALL \
--security-opt no-new-privileges \
-v "$(pwd):/workspace:ro" \
openclaw-sandbox
Profile: Standard (for read/write skills)
FROM node:20-alpine
RUN adduser -D -h /workspace openclaw
WORKDIR /workspace
USER openclaw
docker run --rm \
--network none \
--cap-drop ALL \
--security-opt no-new-privileges \
--memory 512m \
--cpus 1 \
--pids-limit 100 \
-v "$(pwd):/workspace" \
openclaw-sandbox
Profile: Network (for skills needing API access)
FROM node:20-alpine
RUN adduser -D -h /workspace openclaw
WORKDIR /workspace
USER openclaw
docker run --rm \
--cap-drop ALL \
--security-opt no-new-privileges \
--memory 512m \
--cpus 1 \
--pids-limit 100 \
--dns 1.1.1.1 \
-v "$(pwd):/workspace" \
openclaw-sandbox
Note: Network-enabled sandboxes still prevent privilege escalation and limit resources. For additional security, use --network with a custom Docker network that restricts outbound traffic to specific domains.
Configuration Generator
When the user provides a skill's permissions, generate the appropriate sandbox:
Input
Skill: <name>
Permissions: fileRead, fileWrite, network, shell
Output
- Dockerfile — minimal base image, non-root user
- docker run command — with all security flags
- docker-compose.yml — for repeated use
What this file has done since we first saw it
Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.
- 12d ago First seen · 146 lines · 29 tokens per session scan A 9c0d71fe99e9
sandbox-guard is a skill published in the GitHub repository UseAI-pro/openclaw-skills-security (71 stars, last pushed 6mo ago), licensed MIT. It adds 29 tokens to every session and 1,141 once invoked, about $0.0001 per session on Opus 5. A static security scan graded it A with 0 findings. No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-30.
Other skills, from other repositories
security-guide
A security deployment guide for OpenClaw, with instructions in English and Chinese. It covers network exposure, container isolation, credentials, permissions, and audit logging.
clawmoat
Real-time AI agent security scanner. Detects prompt injection, jailbreak attempts, credential/secret leaks, PII exposure, and dangerous tool calls. Activate when: (1) scanning inbound messages or tool outputs for prompt injection, (2) checking outbound content for credential leaks or PII, (3) auditing agent session…
trent-openclaw-security
Assess your Agent deployment against security risks using Trent.
auditing-ecs-task-metadata-boundaries
Audit container task credential and metadata boundaries in orchestrated compute such as ECS: a workload that can reach the container credential endpoint or the host instance metadata service to obtain a role broader than the task needs, a task role over-scoped for the workload, a sidecar or co-located container…
auditing-host-mount-and-device-exposure
Audit the host paths and devices a workload mounts for reach across the container boundary onto the node: a writable hostPath into a sensitive node directory, a mount of the host root or a system path that exposes other pods' data and node configuration, a raw device or block volume that grants low-level host access…
auditing-init-and-sidecar-injection-trust
Audit the init and sidecar containers a workload runs, including ones injected by a mutating admission webhook, for trust the main container never granted: an injected sidecar that runs with broader privileges, host access, or credentials than the workload, an init container that fetches and executes remote content…