← Back to Skills Marketplace
krisco65

agent-wake

by krisco65 · GitHub ↗ · v1.0.1
cross-platform ⚠ suspicious
445
Downloads
0
Stars
4
Active Installs
2
Versions
Install in OpenClaw
/install agent-wake
Description
Wake an OpenClaw agent session from an external script or process. Use when a background task (Claude Code CLI, cron job, webhook, price alert, or any script...
README (SKILL.md)

agent-wake

Wake your OpenClaw agent from any external process using the gateway HTTP API.

How it works

scripts/agent-wake.py calls POST /tools/invoke with the cron tool, firing a wake event into the agent's session. The agent receives the event text as a system message and responds immediately in the correct channel.

Quick start

python agent-wake.py "Task finished -- brief summary" "YOUR_DISCORD_CHANNEL_ID"

Setup (one-time)

See references/setup.md for:

  • Enabling the cron tool over HTTP (required -- blocked by default)
  • Setting GATEWAY_TOKEN
  • Finding your Discord channel ID

Usage patterns

End of a Claude Code CLI task

Add to the task prompt:

When done, run: python "/path/to/agent-wake.py" "Task done -- summary here" "CHANNEL_ID"

From any Python script

import subprocess
subprocess.run([
    "python", "/path/to/agent-wake.py",
    "Price alert triggered -- AAPL crossed $200",
    "1475232925724315740"
])

Standalone (wake main session)

python agent-wake.py "Backup completed successfully"

Omit channel ID to wake the main session (response goes to default channel).

What the agent receives

The event text is injected as a system message. Be specific -- the agent acts on what you write:

Build finished -- 3 errors fixed, tests passing. Send your response to Discord channel 1475232925724315740...

Script location

scripts/agent-wake.py -- copy this wherever your tasks run. No dependencies beyond Python stdlib.

Usage Guidance
The included Python script appears to do exactly what the description says: post a wake event to the local OpenClaw gateway using a gateway auth token. However, the registry metadata incorrectly lists no required credentials while the SKILL.md and script require a GATEWAY_TOKEN (or read OPENCLAW_GATEWAY_TOKEN from ~/.openclaw/gateway.cmd). Before installing or running this: 1) confirm where you should store the gateway token and that the registry metadata is updated; 2) ensure the gateway URL is a trusted, local endpoint (default is http://localhost:18789) and that enabling the cron tool over HTTP is acceptable for your environment; 3) protect the gateway token (do not commit a .env with the token into version control); and 4) if you plan to run this from external or networked environments, consider network-level protections so the gateway cannot be invoked by untrusted actors. If you want to proceed, update the skill/metadata to explicitly declare GATEWAY_TOKEN as a required credential so its needs are clear.
Capability Analysis
Type: OpenClaw Skill Name: agent-wake Version: 1.0.1 The skill is designed to wake an OpenClaw agent session via a local gateway, injecting a user-provided message as a system event. It accesses the `GATEWAY_TOKEN` for authentication (from environment, local .env, or `~/.openclaw/gateway.cmd`) and makes an HTTP POST request to `http://localhost:18789/tools/invoke`. While the core functionality involves injecting text that the AI agent will act upon (a form of prompt injection), this is the explicit and documented purpose of the skill, and the script itself adds instructions to *constrain* the agent's response (e.g., 'Do not respond anywhere else.'), not to manipulate it maliciously. There is no evidence of data exfiltration, unauthorized execution, persistence, or other malicious intent from the skill itself. The `references/setup.md` clearly outlines the required manual configuration to allow the `cron` tool, which is a prerequisite for the skill's operation, not a malicious action.
Capability Assessment
Purpose & Capability
The SKILL.md and the included script require a gateway auth token (GATEWAY_TOKEN or OPENCLAW_GATEWAY_TOKEN read from ~/.openclaw/gateway.cmd), but the registry metadata lists no required environment variables or credentials. That mismatch is incoherent: a wake tool legitimately needs the gateway token, and the metadata should declare it. Other than that, the requested actions (POST to /tools/invoke on the gateway) align with the stated purpose.
Instruction Scope
The runtime instructions and script are narrowly scoped: they load a local .env (script directory only), optionally read gateway.cmd from ~/.openclaw, and POST a wake event to the local gateway /tools/invoke endpoint. The README instructs enabling the cron tool over HTTP and setting the gateway token. The skill does not instruct reading unrelated system files or transmitting data to third-party endpoints.
Install Mechanism
No install spec; this is an instruction-only skill with a small Python script. Nothing is downloaded or written to disk by an installer. Risk from installation is low.
Credentials
The functionality reasonably requires a gateway auth token, and the SKILL.md declares GATEWAY_TOKEN as required, but the registry metadata omitted it. The script also reads OPENCLAW_GATEWAY_TOKEN from gateway.cmd—this is relevant to the purpose but the lack of declared credentials in the registry is a red flag for packaging/metadata correctness and could lead to accidental misconfiguration or unintended exposure.
Persistence & Privilege
The skill does not request permanent presence, does not set always:true, and does not modify other skills or system-wide settings. It merely invokes the gateway API when executed.
How to Use
  1. Make sure OpenClaw is installed (local or Docker)
  2. Run the install command in chat: /install agent-wake
  3. After installation, invoke the skill by name or use /agent-wake
  4. Provide required inputs per the skill's parameter spec and get structured output
Version History
v1.0.1
Fix: removed broad .env scanning; now only reads local .env next to script or gateway token from gateway.cmd. Added explicit credential declaration in metadata.
v1.0.0
Initial release: wake your OpenClaw agent from any external script or Claude Code CLI task via the gateway HTTP API
Metadata
Slug agent-wake
Version 1.0.1
License
All-time Installs 4
Active Installs 4
Total Versions 2
Frequently Asked Questions

What is agent-wake?

Wake an OpenClaw agent session from an external script or process. Use when a background task (Claude Code CLI, cron job, webhook, price alert, or any script... It is an AI Agent Skill for Claude Code / OpenClaw, with 445 downloads so far.

How do I install agent-wake?

Run "/install agent-wake" in the OpenClaw or Claude Code chat to install it in one step — no extra setup required.

Is agent-wake free?

Yes, agent-wake is completely free (open-source). You can download, install and use it at no cost.

Which platforms does agent-wake support?

agent-wake is cross-platform and runs anywhere OpenClaw / Claude Code is available (cross-platform).

Who created agent-wake?

It is built and maintained by krisco65 (@krisco65); the current version is v1.0.1.

💬 Comments