Practical overview
When it fits
Use Axe when agent work should be configured as reusable CLI tasks and composed with existing shell or developer automation. It executes the agent run; an external tool handles scheduling.
Who it is for
- Developers who want reusable, task-specific LLM agents in shell workflows.
- Teams that want to define agent settings in version-controllable TOML files.
- Users who need to pass files, diffs or other command output into agent runs.
Documented uses
- Configure an agent with a model, prompt, skill files and working directory.
- Run a focused agent on data piped from another command.
- Pass results between specialized agents.
- Trigger agent execution from existing automation tools.
- Inspect resolved agent context with dry-run mode or emit structured JSON for scripts.
Requirements & limits
The repository documents pre-built binaries for Linux, macOS and Windows. Building or installing with Go requires Go 1.25 or later.
- Agent definitions are TOML files and can specify prompts, models, skills, context, tools and memory.
- Documented LLM providers include Anthropic, OpenAI, Ollama, OpenCode and AWS Bedrock.
- Axe is an executor, not a scheduler; automation triggers come from external tools.
- The CLI documents sub-agent delegation, JSON output, dry-run mode, token budgets and configurable retries.
- Axe is licensed under Apache-2.0.
Check current requirements in the official documentation. Source-based guidance does not mean OpenViST has run or independently tested this project.
Set up and run a TOML-defined agent
Initialize Axe, create an agent configuration and invoke it from the CLI.
Before you start
- Axe installed; installing or building with Go requires Go 1.25 or later.
- A TOML agent definition and the required LLM provider credentials for a provider-backed run.
Steps
Run
axe config initto create the configuration directory and sample files.Run
axe agents init my-agentto scaffold an agent.Edit the agent configuration with its model and task instructions.
Run
axe run my-agent.
Expected outcome
Axe invokes the named configured agent; provider-backed execution requires credentials for the selected provider.
Guide limitations
- Axe executes agents but does not provide workflow scheduling.
- The documented quick start does not specify a provider setup procedure.
Pass a Git diff to an agent through stdin
Send Git diff output to a configured agent for a focused task such as code review.
Before you start
- Axe installed and a configured agent, such as a code reviewer.
- A Git diff to provide as standard input.
- Credentials for the configured LLM provider when the run calls that provider.
Steps
Create or configure an agent for the desired task.
Pipe the diff into Axe, for example: the command below.
git diff --cached | axe run pr-reviewer
Expected outcome
The agent receives the diff as input and returns its result through the CLI.
Guide limitations
- Axe performs the agent run; scheduling or other workflow orchestration must come from external tools.
- The example requires an agent named
pr-reviewerto be configured.
Sources & review
Source record details
- Source availability
- Active
- Sources checked
- 14 October 2026
- Original website title
- jrswab/axe · Official homepage title · retained in its source language

