Back to skill

Security audit

agent-easy-http

Security checks for vulnerabilities and agentic risk

Overview

ClawScan marks this skill suspicious because scanners report a remotely callable agent interface, persistent service and configuration changes, and several security concerns; the scan is partial, so the evidence is incomplete.

Consider the suspicious verdict and incomplete scan before installing. The reported concerns include remote agent access, credential and prompt exposure over unencrypted connections, an active probe described as read-only, persistent service and hooks changes, and a potentially root-run service. Review these risks and confirm that the network access, elevated permissions, and persistence match your needs.

Vulnerability Patterns
  • Skill Instruction HijackingAlters the agent's session goals or safety constraints when the skill loads
  • Agent Memory PoisoningWrites attacker-controlled rules into memory that affect later sessions
  • Remote Payload Retrieval and ExecutionFetches external code whose behavior can change after review
  • Embedded Malicious CodeShips malicious scripts inside the skill and executes them locally
  • Unauthorized Access and Privilege EscalationObtains permissions beyond the task's legitimate needs
Vulnerability Patterns
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
  • Tool MisuseTool Parameter Abuse, Chaining Abuse, Unsafe Defaults
  • Rogue AgentSelf-Modification, Session Persistence
  • Behavioral ASTexec() Call, eval() Call, Dynamic Import
Findings (48)

Tool Parameter Abuse

High
Category
Tool Misuse
Confidence
60% confidence
Finding

Tool parameters are crafted to achieve unintended or unsafe behavior. Parameter abuse can bypass intended safety checks (e.g. shell=True, --force, dangerous glob patterns).

Content

Scanner excerpt · README.md (reported line 14)May include surrounding context.

md
` is an open-ended entry point.** Any authenticated caller can send **arbitrary prompts** to the agent, which then decides which Skills to run — including side-effecting ones (send messages, modify data). Keep the port reachable only by trusted callers.
- **HTTP is the default and is unencrypted.** By default the `X-API-Key`, prompts, and results travel in **cleartext**. Acceptable only on `localhost` / a trusted network; enable HTTPS when binding `0.0.0.0` or calling cross-host, and never use `curl -k` (skip cert verification) in production.
- **It modifies global config.** init / watchdog write and self-heal the hooks config in `~/.openclaw/openclaw.json` (and optionally sync to external sources via `OPENCLAW_CONFIG_SYNC_PATHS`). This is a platform-level change; the watchdog may re-apply it after a manual revert — be aware of this persistence behavior.
- **Protect the API Key.** It is a bearer credential — leaking it grants agent-invocation rights. Keep it out of shell history, logs,

Tool Parameter Abuse

High
Category
Tool Misuse
Confidence
60% confidence
Finding

Tool parameters are crafted to achieve unintended or unsafe behavior. Parameter abuse can bypass intended safety checks (e.g. shell=True, --force, dangerous glob patterns).

Content

Scanner excerpt · SKILL.md (reported line 28)May include surrounding context.

md
> 本 skill 会启动一个常驻代理,把你的 **OpenClaw agent 暴露成可远程调用的 HTTP 接口**。agent 具备文件、shell、消息发送等本机能力,因此这是一个**对外开放的远程控制面**,请理解以下风险边界:
>
> - **`/agent/run` 是通用入口(v1.1.0 起默认关闭)**:开启后任何通过鉴权的调用方都能向 agent 发送**任意 prompt**,等价于完全交出 agent 控制权。默认 403,需在 config 显式 `enable_agent_run: true`。
> - **默认仅监听本机(v1.1.0 起)**:`listen_host` 默认 `127.0.0.1`。监听 `0.0.0.0` 且 TLS 未启用时**启动直接失败**,需 `AGENT_EASY_HTTP_ALLOW_INSECURE=1` 或 config `allow_insecure_public_bind: true` 显式豁免(跨主机务必启用 HTTPS,切勿生产 `curl -k`)。
> - **默认 HTTP 不加密**:默认模式下 `X-API-Key`、prompt、执行结果均**明文传输**。仅 `localhost` / 可信内网可接受。
> - **会修改全局配置**:init / watchdog 会写入并自愈 `~/.openclaw/openclaw.json` 的 hooks 配置(可选还会同步到 `OPENCLAW_CONFIG_SYNC_PATHS` 外部配置源)。这是平台级改动,管理员手动回退后 watchdog 可能重新应用,请知悉其持久化行为。
> - **保护好 API Key**:Key 是 bearer 凭证,泄露即等同交出 agent 调用权。注意别进入 shell history、日志、截图或权限过宽的文件。

Incomplete referenced artifact analysis

High
Category
analysis-evasion
Confidence
100% confidence
Finding

Referenced artifact was not completely inspected

Content

Scanner excerpt · SKILL.md (reported line 361)May include surrounding context.

md
详细客户端信任指引见 `references/deployment.md`。

Tool Parameter Abuse

High
Category
Tool Misuse
Confidence
60% confidence
Finding

Tool parameters are crafted to achieve unintended or unsafe behavior. Parameter abuse can bypass intended safety checks (e.g. shell=True, --force, dangerous glob patterns).

Content

Scanner excerpt · references/deployment.md (reported line 162)May include surrounding context.

bash
# curl:临时用 --cacert
curl --cacert ~/.http/certs/server.crt -H "X-API-Key: $API_KEY" \
     https://<your-server-ip>:7720/health

# 或导入系统信任库(Linux)

Tool Parameter Abuse

High
Category
Tool Misuse
Confidence
60% confidence
Finding

Tool parameters are crafted to achieve unintended or unsafe behavior. Parameter abuse can bypass intended safety checks (e.g. shell=True, --force, dangerous glob patterns).

Content

Scanner excerpt · references/deployment.md (reported line 162)May include surrounding context.

bash
# curl:临时用 --cacert
curl --cacert ~/.http/certs/server.crt -H "X-API-Key: $API_KEY" \
     https://<your-server-ip>:7720/health

# 或导入系统信任库(Linux)

Tool Parameter Abuse

High
Category
Tool Misuse
Confidence
95% confidence
Finding

The troubleshooting guidance recommends curl -k, which disables server certificate verification and permits an active network attacker to impersonate the API server. The debugging-only qualification limits intended use but does not protect a real API key or sensitive agent traffic sent during that debugging session.

Content

Scanner excerpt · references/deployment.md (reported line 251)May include surrounding context.

md
## 8. 常见问题

### Q1:客户端报「证书不受信任」
A:把 `~/.http/certs/server.crt` 加到客户端 CA 信任链,或临时用 `curl -k` 跳过校验(仅调试)。

### Q2:服务起不来,提示「api_key 不能为空」或「hook_url 无法解析」
A:跑一遍向导:`python3 scripts/server.py init`(会启用 OpenClaw hooks + 生成 API Key)。

Tool Parameter Abuse

High
Category
Tool Misuse
Confidence
96% confidence
Finding

The generated client guidance explicitly offers curl -k as a temporary alternative, disabling server certificate verification. Although this only prints advice, callers applying it to the agent API could expose bearer credentials, prompts, and results to an active network attacker, potentially granting agent invocation rights.

Content

Scanner excerpt · scripts/gen_cert.py (reported line 188)May include surrounding context.

python
print(f"证书路径:{CERT_PATH}")
    print()
    print("【curl】")
    print(f"  curl --cacert {CERT_PATH} <redacted-url>")
    print("  # 或临时跳过验证:curl -k https://...")
    print()
    print("【Python requests】")

Tool Parameter Abuse

High
Category
Tool Misuse
Confidence
96% confidence
Finding

This finding identifies the same printed recommendation to bypass TLS certificate verification. If copied into authenticated agent requests, that recommendation enables server impersonation and credential theft by an active network attacker; it does not itself execute the insecure request.

Content

Scanner excerpt · scripts/gen_cert.py (reported line 188)May include surrounding context.

python
print(f"证书路径:{CERT_PATH}")
    print()
    print("【curl】")
    print(f"  curl --cacert {CERT_PATH} <redacted-url>")
    print("  # 或临时跳过验证:curl -k https://...")
    print()
    print("【Python requests】")

The advertised deny-list protection is bypassed by the optional generic agent endpoint.

High
Category
Excessive Agency
Confidence
94% confidence
Finding

The manifest advertises a deny-list filter, but deny_skills only filters discovery and named skill requests (L0325–L0328 and L0714–L0717). When enable_agent_run is explicitly enabled, /agent/run forwards requests to an unrestricted agent without checking or conveying the deny list, allowing callers to request skills that the proxy otherwise reports as denied.

Content

No source excerpt is available for this finding.

os.system() or os exec-family call

High
Category
Dangerous Code Execution
Confidence
85% confidence
Finding

os.system() and os exec-family calls run shell commands with the process's full privileges, enabling arbitrary command execution.

Content

Scanner excerpt · scripts/server.py (reported line 1076)May include surrounding context.

python
time.sleep(1)
    if args.command == "init":
        wizard = Path(__file__).parent / "init_wizard.py"
        os.execvp(sys.executable, [sys.executable, str(wizard)])
    if args.command == "setup-hooks":
        wizard = Path(__file__).parent / "init_wizard.py"
        os.execvp(sys.executable, [sys.executable, str(wizard), "--setup-hooks-only"])

os.system() or os exec-family call

High
Category
Dangerous Code Execution
Confidence
85% confidence
Finding

os.system() and os exec-family calls run shell commands with the process's full privileges, enabling arbitrary command execution.

Content

Scanner excerpt · scripts/server.py (reported line 1079)May include surrounding context.

python
os.execvp(sys.executable, [sys.executable, str(wizard)])
    if args.command == "setup-hooks":
        wizard = Path(__file__).parent / "init_wizard.py"
        os.execvp(sys.executable, [sys.executable, str(wizard), "--setup-hooks-only"])

    # 生成默认配置 + 加载
    _save_default_config()

Chaining Abuse

High
Category
Tool Misuse
Confidence
75% confidence
Finding

Tool calls are chained to bypass individual safety checks or escalate capabilities beyond what any single tool call would allow.

Content

Scanner excerpt · scripts/watchdog.sh (reported line 186)May include surrounding context.

sh
export WATCHDOG_FOREGROUND=1
    echo "$$" > "$WATCHDOG_PID_FILE"
    log "Watchdog daemon started (interval=${CHECK_INTERVAL}s, log=$LOG_FILE)"
    trap 'log "Watchdog stopping (signal)"; rm -f "$WATCHDOG_PID_FILE"; exit 0' SIGTERM SIGINT
    while true; do
        if ! is_server_healthy; then
            log "Service down or unhealthy, recovering..."

External Transmission

Medium
Category
Data Exfiltration
Confidence
60% confidence
Finding

Data is being sent to an external URL. This could be legitimate telemetry or data exfiltration. Manual review is recommended.

Content

Scanner excerpt · README.md (reported line 14)May include surrounding context.

md
` is an open-ended entry point.** Any authenticated caller can send **arbitrary prompts** to the agent, which then decides which Skills to run — including side-effecting ones (send messages, modify data). Keep the port reachable only by trusted callers.
- **HTTP is the default and is unencrypted.** By default the `X-API-Key`, prompts, and results travel in **cleartext**. Acceptable only on `localhost` / a trusted network; enable HTTPS when binding `0.0.0.0` or calling cross-host, and never use `curl -k` (skip cert verification) in production.
- **It modifies global config.** init / watchdog write and self-heal the hooks config in `~/.openclaw/openclaw.json` (and optionally sync to external sources via `OPENCLAW_CONFIG_SYNC_PATHS`). This is a platform-level change; the watchdog may re-apply it after a manual revert — be aware of this persistence behavior.
- **Protect the API Key.** It is a bearer credential — leaking it grants agent-invocation rights. Keep it out of shell history, logs,

Session Persistence

Medium
Category
Rogue Agent
Confidence
88% confidence
Finding

The documented watchdog can reapply global hooks configuration after an administrator manually reverts it, potentially undoing deliberate revocation of an agent invocation surface. Disclosure makes this less suspicious as malicious persistence, but does not resolve the conflict with administrative security changes.

Content

Scanner excerpt · README.md (reported line 15)May include surrounding context.

md
caller can send **arbitrary prompts** to the agent, which then decides which Skills to run — including side-effecting ones (send messages, modify data). Keep the port reachable only by trusted callers.
- **HTTP is the default and is unencrypted.** By default the `X-API-Key`, prompts, and results travel in **cleartext**. Acceptable only on `localhost` / a trusted network; enable HTTPS when binding `0.0.0.0` or calling cross-host, and never use `curl -k` (skip cert verification) in production.
- **It modifies global config.** init / watchdog write and self-heal the hooks config in `~/.openclaw/openclaw.json` (and optionally sync to external sources via `OPENCLAW_CONFIG_SYNC_PATHS`). This is a platform-level change; the watchdog may re-apply it after a manual revert — be aware of this persistence behavior.
- **Protect the API Key.** It is a bearer credential — leaking it grants agent-invocation rights. Keep it out of shell history, logs, screenshots, and world-readable files.

## Features

Unsafe Defaults

Medium
Category
Tool Misuse
Confidence
70% confidence
Finding

Tool defaults are unsafe or overly permissive (e.g. disabled TLS verification, no authentication, world-writable permissions). Unsafe defaults widen the attack surface.

Content

Scanner excerpt · README.md (reported line 16)May include surrounding context.

md
aller can send **arbitrary prompts** to the agent, which then decides which Skills to run — including side-effecting ones (send messages, modify data). Keep the port reachable only by trusted callers.
- **HTTP is the default and is unencrypted.** By default the `X-API-Key`, prompts, and results travel in **cleartext**. Acceptable only on `localhost` / a trusted network; enable HTTPS when binding `0.0.0.0` or calling cross-host, and never use `curl -k` (skip cert verification) in production.
- **It modifies global config.** init / watchdog write and self-heal the hooks config in `~/.openclaw/openclaw.json` (and optionally sync to external sources via `OPENCLAW_CONFIG_SYNC_PATHS`). This is a platform-level change; the watchdog may re-apply it after a manual revert — be aware of this persistence behavior.
- **Protect the API Key.** It is a bearer credential — leaking it grants agent-invocation rights. Keep it out of shell history, logs, screenshots, and world-readable files.

## Features

Skill declares no tool scope ('permissions' or 'allowed-tools') but code capabilities were detected: env, file_read, file_write, network, shell.

Medium
Category
MCP Least Privilege
Confidence
70% confidence
Finding

Without declared permissions the skill's intent is opaque and cannot be validated.

Content

No source excerpt is available for this finding.

External Transmission

Medium
Category
Data Exfiltration
Confidence
60% confidence
Finding

Data is being sent to an external URL. This could be legitimate telemetry or data exfiltration. Manual review is recommended.

Content

Scanner excerpt · SKILL.md (reported line 153)May include surrounding context.

md
CERT=<workspace>/.http/certs/server.crt

# 跑指定 skill
curl --cacert $CERT -H "X-API-Key: $API_KEY" \
     -H "Content-Type: application/json" \
     -d '{"message":"运行 hello-env"}' \
     https://192.168.1.10:7720/skills/hello-env/run

Sudo/Root Execution

Medium
Category
Privilege Escalation
Confidence
70% confidence
Finding

Commands invoke sudo or root privileges. Verify this elevated access is necessary and justified.

Content

Scanner excerpt · SKILL.md (reported line 357)May include surrounding context.

curl --cacert /.http/certs/server.crt -H "X-API-Key: $KEY" ...

方式 B:导入系统信任库(Linux 示例)

sudo cp /.http/certs/server.crt /usr/local/share/ca-certificates/ sudo update-ca-certificates

text

Sudo/Root Execution

Medium
Category
Privilege Escalation
Confidence
70% confidence
Finding

Commands invoke sudo or root privileges. Verify this elevated access is necessary and justified.

Content

Scanner excerpt · SKILL.md (reported line 358)May include surrounding context.

方式 B:导入系统信任库(Linux 示例)

sudo cp /.http/certs/server.crt /usr/local/share/ca-certificates/ sudo update-ca-certificates

text

详细客户端信任指引见 `references/deployment.md`。

External Transmission

Medium
Category
Data Exfiltration
Confidence
99% confidence
Finding

Although the unauthenticated health request is harmless, the surrounding curl example transmits the API key, agent requests, and results over plaintext HTTP. The documented default of listening on all interfaces increases exposure, and possession of the intercepted key grants access to an agent execution API.

Content

Scanner excerpt · references/deployment.md (reported line 65)May include surrounding context.

md
BASE=http://<your-server-ip>:7720

# 健康检查(无需鉴权)
curl $BASE/health

# 列出可用 skill
curl -H "X-API-Key: $API_KEY" $BASE/skills

External Transmission

Medium
Category
Data Exfiltration
Confidence
99% confidence
Finding

The Python example sends the API key and agent prompt over plaintext HTTP to a network host. An attacker able to observe or intercept that traffic could steal the key, read or modify prompts, and invoke the exposed agent with the stolen credential; the intended destination does not make this malicious exfiltration, but transport protection is missing.

Content

Scanner excerpt · references/deployment.md (reported line 98)May include surrounding context.

md
HEADERS = {"X-API-Key": API_KEY}

# 触发
resp = requests.post(
    f"{BASE}/agent/run",
    headers=HEADERS,
    json={"message": "帮我生成本周工作总结"},

External Transmission

Medium
Category
Data Exfiltration
Confidence
60% confidence
Finding

Data is being sent to an external URL. This could be legitimate telemetry or data exfiltration. Manual review is recommended.

Content

Scanner excerpt · references/deployment.md (reported line 162)May include surrounding context.

bash
# curl:临时用 --cacert
curl --cacert ~/.http/certs/server.crt -H "X-API-Key: $API_KEY" \
     https://<your-server-ip>:7720/health

# 或导入系统信任库(Linux)

Sudo/Root Execution

Medium
Category
Privilege Escalation
Confidence
70% confidence
Finding

Commands invoke sudo or root privileges. Verify this elevated access is necessary and justified.

Content

Scanner excerpt · references/deployment.md (reported line 166)May include surrounding context.

https://:7720/health

或导入系统信任库(Linux)

sudo cp ~/.http/certs/server.crt /usr/local/share/ca-certificates/agent-easy-http.crt sudo update-ca-certificates

text

Sudo/Root Execution

Medium
Category
Privilege Escalation
Confidence
70% confidence
Finding

Commands invoke sudo or root privileges. Verify this elevated access is necessary and justified.

Content

Scanner excerpt · references/deployment.md (reported line 167)May include surrounding context.

或导入系统信任库(Linux)

sudo cp ~/.http/certs/server.crt /usr/local/share/ca-certificates/agent-easy-http.crt sudo update-ca-certificates

text

浏览器信任自签证书:

Sudo/Root Execution

Medium
Category
Privilege Escalation
Confidence
70% confidence
Finding

Commands invoke sudo or root privileges. Verify this elevated access is necessary and justified.

Content

Scanner excerpt · references/deployment.md (reported line 215)May include surrounding context.

启用:

bash
sudo systemctl daemon-reload
sudo systemctl enable agent-easy-http
sudo systemctl start agent-easy-http
sudo journalctl -u agent-easy-http -f