{% extends "base.html" %}
{% block title %}Quickstart | {{ app_name }}{% endblock %}
{% block body_class %}landing{% endblock %}
{% block body %}
<header><div class="wrap">
  <a class="brand" href="/"><span class="brand-mark" aria-hidden="true"></span>{{ app_name }}</a>
  <nav aria-label="Main">{% if session %}<a class="btn" href="/app">Open dashboard</a>{% else %}<a href="/login">Log in</a>{% endif %}</nav>
</div></header>
<main class="wrap prose doc">
  <h1>Quickstart</h1>
  <p class="muted">Connect an agent in about ten minutes. The examples use Gemini; OpenAI and Anthropic are further down.</p>

  <h2>1. Get an API key</h2>
  <p>{% if session %}<a href="/app/keys">Create a key</a>{% else %}<a href="/signup">Create an account</a>, then create a key{% endif %} and set it in your environment:</p>
<pre class="snippet">pip install agent-supra
pip install google-genai
export AGENT_SUPRA_API_KEY="as_live_…"
export AGENT_SUPRA_URL="{{ base_url }}"</pre>

  <h2>2. Wrap your agent</h2>
  <p>Create a task with a budget, wrap your model client, and report each test run. <code>rev</code> is anything that changes when the code changes, such as a git commit hash or a hash of the edited files.</p>
<pre class="snippet">from google import genai
from agent_supra import Client, SupervisedRun, SupervisorPaused

sup = Client()
task = sup.create_task("Fix failing checkout tests", budget_usd=3)

with SupervisedRun(sup, task) as run:
    gemini = run.wrap_gemini(genai.Client())
    try:
        while True:
            run.check()                       # stops here once paused
            reply = gemini.models.generate_content(
                model="gemini-3.5-flash-lite", contents=prompt)
            run.tool("edit_file", apply_fix, reply.text, rev=git_rev())
            result = run.tool("run_tests", run_tests, rev=git_rev())
            run.report_tests(result.all_ids, result.failing_ids, rev=git_rev())
            if not result.failing_ids:
                run.finish("success")
                break
    except SupervisorPaused as paused:
        print("Paused:", paused.reason)       # don't retry; review it in the dashboard</pre>

  <h2>3. Watch it in observation mode</h2>
  <p>New workspaces start in observation mode: runs are flagged, never paused, while you check the flags are right. Budgets and call caps are enforced from the start. When you trust the rules, turn observation mode off in <a href="/app/settings">Settings</a> and add a Slack webhook so you hear about pauses.</p>

  <h2>Other model clients</h2>
<pre class="snippet"># OpenAI (also works for Gemini's OpenAI-compatible endpoint)
client = run.wrap_openai(openai.OpenAI())
client.chat.completions.create(model="gpt-…", messages=[…])

# Anthropic
client = run.wrap_anthropic(anthropic.Anthropic())
client.messages.create(model="claude-…", max_tokens=1024, messages=[…])

# Anything else: record usage yourself
run.record_model_call("my-model", input_tokens=1200, output_tokens=300)</pre>
  <p>Gemini prices are built in. For other models, add prices under <a href="/app/settings">Settings</a> so spend can be estimated.</p>

  <h2>Sub-agents</h2>
  <p><code>child = run.child()</code> creates a run that shares the parent task's budget.</p>

  <h2>If the service is unreachable</h2>
  <p>By default the SDK fails open: your agent keeps working, events are kept in memory and sent when the service is back. Use <code>Client(fail_mode="closed")</code> to stop the agent instead when supervision isn't available.</p>

  <h2>What gets sent</h2>
  <p>Token counts, model names, test IDs with pass or fail, code revision IDs, and keyed fingerprints of tool arguments and errors. Prompts, replies and code are never sent. See <a href="/privacy">Privacy</a>.</p>
</main>
{% endblock %}