Give it a purpose.
Choose a task and a success check. Set the harness, model, environment, and access.
✳ Happy paths. Better products.
A good agent experience
is built on purpose.
Your agent has a goal. Your product has a part to play. AJX helps you see where the two get lost in translation.
One prompt installs AJX: the CLI and the skill, together.
A small task. An interesting journey.Fictional example
→
“Export 100 records to CSV.”
A clear goal, with a result we can verify: 100 data rows in a valid CSV.
⑂
Export or dump? The first detour.
Both commands look relevant. The top-level help does not explain which one writes CSV.
!
The quick start has a wrong turn.
The documented --out flag fails. A help lookup reveals the supported --output flag.
✓
A happy ending. Three useful asks.
The task succeeds. The detours still reveal three ways to improve the product.
Pick a stop. Follow the evidence.
⑂ Follow the forks! Find the friction✓ Verify the result↗ Improve the next attempt
A fresh agent. A useful perspective.
Your product has a job to do.
Let’s see how it helps.
AJX gives a fresh worker a real task, records its journey, and turns the evidence into product improvements your team can act on.
See how it goes.
Follow the attempt in your local console. Keep the events and artifacts, including the detours.
Make the next try better.
Open the report. Inspect the asks. Follow each one back to the evidence before deciding what to fix.
A happy ending isn’t the whole story.
The task worked.
Here’s what could work better.
A concrete change. The reason it matters. A way to check it. That’s what an AJX product ask gives your team.
Fictional sample · Demo Export CLI
01!Roadblock
Make the quick-start command executable.
The docs use --out. The CLI expects --output. The first attempt fails before the agent finds the correction.
What to change
Use the supported flag in a copyable CSV example.
How to check
The documented command creates a valid 100-row CSV.
02◷Wait
Show progress while the export runs.
Only the start and completion are visible during a 47-second export. The agent cannot tell working from stuck.
What to change
Emit progress separately from the CSV output.
How to check
A delayed export remains observable without polluting stdout.
03⑂Fork
Make the first choice a clear one.
Both export and dump sound right. Top-level help does not explain which command writes CSV.
What to change
Explain formats and intended use in command help.
How to check
An agent can identify the CSV command from top-level help.
⑂
Walk the full journey
↗▦
Forks, blocks, waits, and helpful moments.
Compare the matrix
↗
Asks and outcomes across versions and harnesses.
These events and measurements are invented to demonstrate the report. No real product or agent was tested in this example.
Your product’s turn
A good first step.
On purpose.
Open your project in a skill-capable agent and paste the prompt. It installs AJX and helps you plan one small, useful attempt.
- ✓ Use macOS or Linux.
- ✓ Install and sign in to a supported worker CLI.
- ✓ Pick a task—or let your agent help you choose.
Before the worker runs
Review the task, success check, access, and expected model costs. Then open the report and visual journey together.
Codex · Claude Code · Kiro CLI
AJX skills work with compatible Agent Skills hosts.
Paste this into your agent
Install the latest stable AJX CLI and skill by following https://raw.githubusercontent.com/royosherove/ajx/main/docs/AGENT-QUICKSTART.md Then load the installed AJX skill. I want to use AJX to test jq. Starting point: https://jqlang.org/manual/ Task: Filter a sample JSON inventory to in-stock items and export their names and prices as CSV. Prepare any tools, generated sample data, and local test server needed in a temporary workspace. Use the AJX skill to run one trial with a fresh agent, verify the result, and show me the report and journey.
Try an example. No project needed.CLI + skill installed together
Prefer a terminal? Install the CLI + skill yourself.
bash -o pipefail -c 'curl -fsSL https://github.com/royosherove/ajx/releases/latest/download/install.sh | bash' && export PATH="$HOME/.local/bin:$PATH"
macOS or Linux. The installer includes the skill. Your agent can read it directly; start a fresh session if the host needs one to discover /ajx. No GitHub login required.
Native CLI + Agent Skill. No Go toolchain or GitHub login.
Start with one attempt;
add matrix runs or CI checks when ready.