use-acpx
Provides knowledge about acpx CLI for agent-to-agent communication. Use when user asks about acpx commands, ACP protocol, agent sessions, prompt queueing, or scriptable agent workflows.
What this skill does
# acpx
## CRITICAL: Background Execution Required
**MUST** use `run_in_background: true` for ALL `acpx` bash commands. Running `acpx` in the foreground will block Claude Code and make the session unusable.
When executing any `acpx` command:
```json
{
"command": "acpx codex 'your prompt here'",
"run_in_background": true
}
```
Failure to run in background will cause Claude Code to block indefinitely while waiting for the agent to complete.
## What acpx is
`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.
**WARNING**: Never use Claude-related ACP agents with acpx. Using Claude to call Claude creates recursive loops and unpredictable behavior. Use alternative agents like `codex`, `openclaw`, `gemini`, etc.
Core capabilities:
- Persistent multi-turn sessions per repo/cwd
- One-shot execution mode (`exec`)
- Named parallel sessions (`-s/--session`, `prompt` verb only)
- Queue-aware prompt submission with optional fire-and-forget (`--no-wait`)
- Cooperative cancel command (`cancel`) for in-flight turns
- Graceful cancellation via ACP `session/cancel` on interrupt
- Session control methods (`set-mode`, `set <key> <value>`)
- Agent reconnect/resume after dead subprocess detection
- Prompt input via stdin or `--file`
- Config files with global+project merge and `config show|init`
- Session metadata/history inspection (`sessions show`, `sessions history`)
- Local agent process checks via `status`
- Stable ACP client methods for filesystem and terminal requests
- Stable ACP `authenticate` handshake via env/config credentials
- Structured streaming output (`text`, `json`, `quiet`)
- Built-in agent registry plus raw `--agent` escape hatch
## Install
```bash
npm i -g acpx
```
For normal session reuse, prefer a global install over `npx`.
## Command model
`prompt` is the default verb. Commands follow this shape:
```bash
acpx [global_options] [<agent>] [verb] [prompt_options] [prompt_text...]
```
Verbs: `prompt` (default), `exec`, `cancel`, `set-mode`, `set`, `status`, `sessions`, `config`. When prompt text is omitted and stdin is piped, `acpx` reads from stdin. For the full grammar, see `./references/cli.md`.
## Built-in agent registry
**IMPORTANT**: Do NOT use Claude-related ACP agents (e.g., `claude`, `@zed-industries/claude-agent-acp`). These create recursive agent-calling-agent scenarios that can cause infinite loops or unexpected behavior.
Friendly agent names resolve to commands:
- `pi` -> `npx pi-acp`
- `openclaw` -> `openclaw acp`
- `codex` -> `npx @zed-industries/codex-acp`
- `gemini` -> `gemini --acp`
- `cursor` -> `cursor-agent acp`
- `copilot` -> `copilot --acp --stdio`
- `droid` -> `droid exec --output-format acp`
- `kimi` -> `kimi acp`
- `opencode` -> `npx -y opencode-ai acp`
- `kiro` -> `kiro-cli acp`
- `kilocode` -> `npx -y @kilocode/cli acp`
- `qwen` -> `qwen --acp`
Rules:
- Default agent is `codex` for top-level `prompt`, `exec`, and `sessions`.
- Unknown positional agent tokens are treated as raw agent commands.
- `--agent <command>` explicitly sets a raw ACP adapter command.
- Do not combine a positional agent and `--agent` in the same command.
## Commands
### Prompt (default, persistent session)
```bash
acpx codex 'fix flaky tests' # implicit prompt
acpx codex prompt 'fix flaky tests' # explicit
acpx prompt 'fix flaky tests' # defaults to codex
```
Auto-resumes the saved session for scope `(agentCommand, cwd, name?)`. Exits code `4` if no session exists — run `sessions new` first. Queue-aware: submits to running queue owner when a prompt is already in flight. On interrupt, sends ACP `session/cancel` before force-kill.
Prompt options: `-s/--session <name>`, `--no-wait`, `-f/--file <path>`.
### Exec (one-shot)
```bash
acpx codex exec 'summarize this repo'
```
Runs a single prompt in a temporary session; does not reuse or persist session state.
**Note**: `-s/--session` is a `prompt`-verb option only — it does not apply to `exec`. For parallel one-shot patterns across separate working dirs, see the Cross-repo / worktree section below.
### Cancel / Mode / Set
```bash
acpx codex cancel
acpx codex set-mode auto
acpx codex set approval_policy conservative
```
`cancel` sends cooperative `session/cancel` via IPC. `set-mode` calls ACP `session/set_mode` (mode ids are adapter-defined). `set` calls ACP `session/set_config_option`. Both route through queue-owner IPC when active.
### Sessions
```bash
acpx sessions list
acpx sessions new [--name <name>]
acpx sessions close [name]
acpx sessions show [name]
acpx sessions history [name] [--limit <count>]
acpx status
acpx codex sessions new --name backend
acpx codex sessions close backend
acpx codex sessions show backend
acpx codex sessions history backend --limit 20
acpx codex status
```
Key behaviors: `sessions` and `sessions list` are equivalent. `new` creates a fresh session for the current scope; if an open session already exists in that scope, it is soft-closed. `show [name]` prints metadata; `history [name]` prints turn previews (default 20).
For the complete subcommand list including `ensure`, all flags, global options, config keys, session scoping rules, queueing internals, output format details, permission modes, and exit codes, see `./references/cli.md`.
## Global options (summary)
- `--agent <command>`: raw ACP command (escape hatch)
- `--cwd <dir>`: working directory for session scope
- `--approve-all` / `--approve-reads` (default) / `--deny-all`: permission mode (mutually exclusive)
- `--format <fmt>`: `text` (default), `json`, `quiet`
- `--timeout <seconds>` / `--ttl <seconds>`: timing controls (TTL default 300s)
- `--verbose`: ACP/debug logs to stderr
## Reference summary
**Config files**: merged in order — `~/.acpx/config.json` (global), `<cwd>/.acpxrc.json` (project). Key settings: `defaultAgent`, `defaultPermissions`, `ttl`, `timeout`, `format`, `agents` map, `auth` map. Run `acpx config show` to inspect; `acpx config init` to create global template.
**Session behavior**: sessions scoped by `agentCommand` + absolute `cwd` + optional `name`. Records in `~/.acpx/sessions/*.json`. Prompt mode auto-resumes by walking up to git root; exits code `4` if no session found (run `sessions new`). Dead PIDs are respawned. Use `-s/--session <name>` for parallel conversations.
**Prompt queueing**: the first `acpx` process for a session becomes the queue owner; others submit over local IPC. Default: wait for completion. `--no-wait`: return after enqueue. `Ctrl+C` sends ACP `session/cancel` before force-kill. Owner shuts down after TTL idle (default 300s).
**Output formats**: `--format text` (default) human-readable stream, `--format json` NDJSON for automation, `--format quiet` final text only, `--json-strict` suppresses non-JSON stderr.
**Permission modes**: choose one — `--approve-all`, `--approve-reads` (default), `--deny-all`. If all requests are denied/cancelled and none approved, `acpx` exits with permission-denied status.
For full details on all options, subcommands, exit codes, and IPC internals, see `./references/cli.md`.
## Practical workflows
Persistent repo assistant (multi-turn):
```bash
acpx codex 'inspect failing tests and propose a fix plan'
acpx codex 'apply the smallest safe fix and run tests'
```
Parallel named sessions + queued follow-up:
```bash
acpx codex -s backend 'fix API pagination bug'
acpx codex -s docs 'draft changelog entry for release'
acpx codex --no-wait 'after tests, summarize root causes and next steps'
```
One-shot and automation:
```bash
acpx --format quiet exec 'summarize repo purpose in 3 lines'
acpx --format json codex 'review current branch changes' > events.ndjson
acpx --cwd ~/repos/shop --approve-all codex -s pr-842 \
'review PR #842 for regressions and propose minimal patch'
```
Cross-repo / worktree delegation (current repo stays untouched):
```bash
# SamRelated in AI Agents
skill-development
IncludedComprehensive meta-skill for creating, managing, validating, auditing, and distributing Claude Code skills and slash commands (unified in v2.1.3+). Provides skill templates, creation workflows, validation patterns, audit checklists, naming conventions, YAML frontmatter guidance, progressive disclosure examples, and best practices lookup. Use when creating new skills, validating existing skills, auditing skill quality, understanding skill architecture, needing skill templates, learning about YAML frontmatter requirements, progressive disclosure patterns, tool restrictions (allowed-tools), skill composition, skill naming conventions, troubleshooting skill activation issues, creating custom slash commands, configuring command frontmatter, using command arguments ($ARGUMENTS, $1, $2), bash execution in commands, file references in commands, command namespacing, plugin commands, MCP slash commands, Skill tool configuration, or deciding between skills vs slash commands. Delegates to docs-management skill for official documentation.
reprompter
IncludedTransform messy prompts into well-structured, effective prompts — single or multi-agent. Use when: "reprompt", "reprompt this", "clean up this prompt", "structure my prompt", rough text needing XML tags and best practices, "reprompter teams", "repromptception", "run with quality", "smart run", "smart agents", multi-agent tasks, audits, parallel work, anything going to agent teams. Don't use when: simple Q&A, pure chat, immediate execution-only tasks. See "Don't Use When" section for details. Outputs: Structured XML/Markdown prompt, quality score (before/after), optional team brief + per-agent sub-prompts, agent team output files. Success criteria: Single mode quality score ≥ 7/10; Repromptception per-agent prompt quality score 8+/10; all required sections present, actionable and specific.
adaptive-compaction
IncludedAdaptive add-on policy and recovery layer that decides WHEN to compact, prune, snapshot, or fork -- replacing fixed-percent auto-compaction across Claude Code, Codex, and MCP-capable hosts. Trigger on auto-compact timing or damage: "when should I compact", "is it safe to compact now or start a fresh session", "auto-compact fires too early/mid-task", "switching to an unrelated task but the window still has space", "context rot", "answers get worse the longer the session runs", "the agent forgot the plan or my decisions after it summarized", "add a layer on top that manages context without changing the agent", raising autoCompactWindow to give the policy room, or installing/tuning a cross-tool compaction policy or PreCompact hook -- even when "compaction" is never said but the problem is context-window pressure or post-summarization memory loss. Do NOT use to summarize a conversation, build RAG, write a summarization prompt (decides WHEN not HOW), or answer max-context-length trivia.
agent-skill-creator
IncludedCreate cross-platform agent skills from workflow descriptions. Activates when users ask to create an agent, automate a repetitive workflow, create a custom skill, or need advanced agent creation. Triggers on phrases like create agent for, automate workflow, create skill for, every day I have to, daily I need to, turn process into agent, need to automate, create a cross-platform skill, validate this skill, export this skill, migrate this skill. Supports single skills, multi-agent suites, transcript processing, template-based creation, interactive configuration, cross-platform export, and spec validation.
llm-wiki
IncludedUse when building or maintaining a persistent personal knowledge base (second brain) in Obsidian where an LLM incrementally ingests sources, updates entity/concept pages, maintains cross-references, and keeps a synthesis current. Triggers include "second brain", "Obsidian wiki", "personal knowledge management", "ingest this paper/article/book", "build a research wiki", "compound knowledge", "Memex", or whenever the user wants knowledge to accumulate across sessions instead of being re-derived by RAG on every query.
skill-master
IncludedAgent Skills authoring, evaluation, and optimization. Create, edit, validate, benchmark, and improve skills following the agentskills.io specification. Use when designing SKILL.md files, structuring skill folders (references, scripts, assets), ingesting external documentation into skills, running trigger evals, benchmarking skill quality, optimizing descriptions, or performing blind A/B comparisons. Keywords: agentskills.io, SKILL.md, skill authoring, eval, benchmark, trigger optimization.