Install to first task

最后更新：2026-11-04

Aiglade is a Windows desktop application that runs several command-line AI agents at once — Claude Code, Codex CLI, Hermes, OpenClaw — inside one window, with a shared task queue and a single approval gate for consequential actions.

Last updated: 4 November 2026

Aiglade doesn't produce the intelligence itself — it's a dispatch console. The work is done by command-line agents already installed on your machine. So the first step is confirming those exist.

How do I use Aiglade without programming? Step 1: check the agents run on their own

On launch, Aiglade detects which agents are available locally and lists them. It finds them by itself; you don't point it at paths.

One practical tip: run one agent in a terminal first and confirm it answers, before using it through Aiglade. Otherwise, if something fails inside Aiglade, you can't tell whether the problem is Aiglade or the agent's own setup.

Why should I get one agent running in Aiglade before thinking about orchestration?

Get one agent running cleanly before you think about orchestration, so environment problems do not get mixed up with task problems. Once an orchestrator is involved, a failure can come from more than one place. The task itself may be unclear, or one of the agents underneath may not be set up properly. Clearing the single-agent case first removes one whole class of causes, which makes later debugging much narrower.

Three checks tell you whether a single agent is really working. It finishes a full round without stopping for environment reasons. It exits normally, without an error exit. Its output lands in the expected directory, opens normally, and matches what the task asked for. When all three hold, the agent is ready to take part in orchestrated work.

Skipping this step tends to waste time in the wrong place. You may rewrite the task several times while it keeps failing, and the task looks like the problem, when something in the environment is actually missing. The other way round, once the environment is fine, problems with the task itself become visible, and fixing the task becomes targeted instead of guesswork.

What is a good first task to dispatch in Aiglade?

Don't test with a real project the first time. Pick an unimportant directory and give it a small, clear task.

Avoid "refactor this module" as a first task — the bigger the goal, the harder it is to judge whether it went well.

How should I phrase my first task in Aiglade so the agent does not have to guess?

The more specific the task, the less the agent has to guess, and the less rework you get. A task description can be split into three sentences, each doing one job. The goal sentence says what should be produced. The input sentence says what material to work from. The acceptance sentence says what counts as done. With all three in place, the agent does not have to guess your intent, and you do not have to judge the result by feel.

A reusable template looks like this:

• Goal: turn these meeting notes into an action list.
• Input: the full text of the meeting notes.
• Acceptance: every action has an owner and a due time, and the list can be handed out as it stands.

Good example one: the goal is to revise a draft into a publishable version, the input is the full draft, and acceptance is that every change is marked while the original structure stays intact.

Good example two: the goal is to file a batch of documents by topic, the input is the folder waiting to be filed, and acceptance is that every document has a topic and can be found by that topic afterwards.

Bad example one: asking it to "improve this project". Nothing says which part to improve or what the result should look like, so the agent can only guess, and a wrong guess means starting over.

Bad example two: asking it to "make this a bit better". Better has no standard here, so at acceptance time nobody can say whether the work is done, and several rounds of revision may still not settle it.

What should I watch during my first full run in Aiglade?

Stay with it the first time. Watch how it breaks the task down and picks actions, and you'll get a feel for its temperament: does it read before editing, or start changing things immediately? When stuck, does it stop and ask, or try another direction on its own?

That observation is worth more than any documentation. Agents differ noticeably — the same task might draw caution from one and recklessness from another.

After a run in Aiglade, how do I check the process, the output, and the record?

A run is only done when the process, the output, and the record all line up. Checking only the output can hide problems in the process. Checking only the process can hide flaws in the output itself. Checking the three separately lets you locate which part went wrong.

Check the process: did the agent follow the expected steps to the end, and did it stop anywhere waiting for a human decision? A stop is not always a failure. Some stops are normal waits for approval, and the approval page explains how those work.

Check the output: did the files land in the expected directory, and is the content complete? Files in the wrong place cost time later. Missing content may mean a step did not run to the end, or the input itself was incomplete.

Check the record: does every step leave a trace, and can a problem be followed back to its cause through that trace? When the record does not match the process, treat the run itself as questionable and do not rush to retry. If the run turns out to be stuck, the task stuck page helps you tell what kind of stall it is before you act.

Where do Aiglade beginners most often get stuck?

Aiglade cannot detect an installed agent: what do I check?

Usually it's installed somewhere Aiglade doesn't look, or under a different user. Check whether the command runs in your terminal under the current user — if it does, it's on PATH.

My Aiglade task shows nothing happening: is it stuck or waiting for approval?

First check whether it's waiting for approval. Consequential actions are held, and a pending item appears in the interface. That's the design working, not a hang.

Is it normal for two agents to give different results for one task in Aiglade?

Expected. Different agents use different models and organise context differently, so the same brief gets read differently. That's an argument for running several: you get two proposals to compare.

If Aiglade does not run after install, what do I check first?

Do not reinstall first. Check in order, and stop at the step that fails, because that step holds the cause. First, can the agent run on its own in a terminal? If it cannot, fix the agent's own installation and setup before touching anything in Aiglade. Second, can Aiglade detect it? If not, it is usually installed somewhere the current user cannot find, or installed under a different user. Third, look at the state of a dispatched task: is it waiting for your approval, or is it genuinely stuck? Waiting for approval is the design working; a genuinely stuck run shows in the process record, at the step where it stopped.

Checking in this order removes one class of causes at a time. Rewriting the task description before the earlier steps pass, or reinstalling before knowing which step fails, usually costs more time than the ordered check.

After my first task in Aiglade, what should I read next?

• Splitting work across agents — which tasks parallelise cleanly
• What the approval gate is for — what gets held, and how to stop it nagging
• Attaching a key you already have — the four common failures whose errors don't name the cause
• Checking that it actually finished — three silent false successes; "no error" proves nothing
• When a task stalls or fails — running, waiting, or already dead