$ loading_
Run headless agent communication, sessions, and multi-agent workflow orchestration.
This skill material is marked as prompt-only and declares no secrets or remote endpoints, so the overall risk is low. The README describes session, filesystem, and terminal capabilities of the local `acpx` CLI, but based on the provided material these appear to be properties of the referenced tool rather than permissions inherently granted by the skill itself.
The header explicitly states that no keys or environment variables are required. While the README mentions authenticate via env/config, the skill itself does not appear to request, store, or broker credentials, and no explicit credential collection or misuse path is described.
The metadata and objective checks both indicate no remote host. Although the description mentions agent-to-agent communication and ACP, it does not identify any concrete external endpoint, third-party service, or user-data egress destination.
The object is explicitly classified as prompt-only; as skill material, it contains no executable script, install hook, or direct local process launcher. The README references `exec`, terminal access, and local agent process checks, but those are capabilities of the referenced CLI rather than proof that this skill itself has code-execution privileges.
No file paths, databases, browser scopes, or system-resource permissions are declared. The README mentions session export/import, history, stdin/`--file`, and filesystem requests, but these are optional capabilities of the external CLI; the skill material itself does not show any granted data-access scope.
The GitHub source is open-source and auditable, which is a positive signal. However, the license is unspecified, community adoption is 0 stars, maintenance status is unknown, and the README suggests global installation via `npm i -g acpx`; this makes the supply-chain maturity and maintenance posture relatively weak, so manual verification of the repository and published package is advisable.
Copy the install command and let the AI configure it · recommended for beginners
Please install the "acpx" skill from askskill: 1. Download https://raw.githubusercontent.com/openclaw/acpx/main/skills/acpx/SKILL.md 2. Save it as ~/.claude/skills/acpx/SKILL.md 3. Reload skills and tell me it's ready
Using acpx, design a CLI workflow that first prompts a research agent to gather information, then passes the result to a writing agent to create a summary. Explain how to configure session scope, output format, and permissions.
A clear acpx usage plan with agent call order, session configuration, permission settings, and example commands.
Using acpx defineFlow, decision, and decisionEdge, design a multi-agent flow where coding questions go to a developer agent and other questions go to a research agent. Output the flow structure and pseudocode.
A clearly executable multi-agent routing design with decision logic, node relationships, and flow pseudocode.
I want to use acpx to process multiple agent tasks in batch. Explain how to set up task queues, isolate different sessions, override the system prompt, and export outputs for both logging and machine parsing.
A batch task management guide covering queueing, session isolation, system prompt overrides, and recommended output formats.
Use this skill when you need to run coding agents through acpx, manage persistent ACP sessions, queue prompts, override the Claude system prompt, prune stale sessions, consume structured agent output from scripts, or compose multi-agent workflows declaratively with acpx/flows.
acpx is a headless, scriptable CLI client for the Agent Client Protocol (ACP). It is built for agent-to-agent communication over the command line and avoids PTY scraping.
Core capabilities:
exec)-s/--session)sessions ensure)sessions prune with age filters and history cleanup)--no-wait)cancel) for in-flight turnssession/cancel on interruptset-mode, set <key> <value>)--fileconfig show|initsessions show, sessions history)statusauthenticate handshake via env/config credentialstext, json, quiet) with optional --suppress-reads--agent escape hatch--system-prompt / --append-system-prompt--no-terminal for review-only flows--allowed-tools), turn cap (--max-turns), retry on transient failures (--prompt-retries)acpx flow run and the acpx/flows authoring API (defineFlow, decision, decisionEdge, acp, action, compute, checkpoint)npm i -g acpx
For normal session reuse, prefer a global install over npx.
prompt is the default verb.
acpx [global_options] [prompt_text...]
acpx [global_options] prompt [prompt_options] [prompt_text...]
acpx [global_options] exec [prompt_options] [prompt_text...]
acpx [global_options] cancel [-s <name>]
acpx [global_options] set-mode <mode> [-s <name>]
acpx [global_options] set <key> <value> [-s <name>]
acpx [global_options] status [-s <name>]
acpx [global_options] sessions [list | new [--name <name>] | ensure [--name <name>] | close [name] | show [name] | history [name] [--limit <count>] | export [name] --output <path> | import <archive> [--name <name>] [--cwd <dir>] | prune [--dry-run] [--before <date> | --older-than <days>] [--include-history]]
acpx [global_options] config [show | init]
acpx [global_options] flow run <file> [--input-json '<json>' | --input-file <path>] [--default-agent <name>]
acpx [global_options] <agent> [prompt_options] [prompt_text...]
acpx [global_options] <agent> prompt [prompt_options] [prompt_text...]
acpx [global_options] <agent> exec [prompt_options] [prompt_text...]
acpx [global_options] <agent> cancel [-s <name>]
acpx [global_options] <agent> set-mode <mode> [-s <name>]
acpx [global_options] <agent> set <key> <value> [-s <name>]
acpx [global_options] <agent> status [-s <name>]
acpx [global_options] <agent> sessions [list | new [--name <name>] | ensure [--name <name>] | close [name] | show [name] | history [name] [--limit <count>] | export [name] --output <path> | import <archive> [--name <name>] [--cwd <dir>] | prune [--dry-run] [--before <date> | --older-than <days>] [--include-history]]
If prompt text is omitted and stdin is piped, acpx reads prompt text from stdin.
Friendly agent names resolve to commands:
…
Fetch GitHub issues, create fixes, open PRs, and handle reviews.
Convert text to speech locally and offline with sherpa-onnx, no cloud needed.
Regenerate OpenClaw release changelog sections from Git history before releases.
Prepare and verify OpenClaw stable or beta releases and release notes.
Automate web page workflows, login checks, tab handling, and recovery steps.
Verify an OpenClaw release is fully published and working across all channels.
Route coding-agent requests into the right ACP or acpx session flow.
Coordinate multiple AI agents safely on the same codebase in real time.
Run web accessibility audits and get compact, actionable fix recommendations.
Manage agent identities, messaging, and shared projects in a shared world.
Connect multiple agent payment protocols for unified charging and settlement workflows.
Build AI agent workflows and automate tasks using MCP-connected services.