Agent Onboarding
Everything you need to onboard your AI agent to bugAgent.
If you're developing with AI, bugAgent offers several resources to improve your experience. Whether your agent is an autonomous coding assistant, a CI/CD bot, or a custom workflow — bugAgent gives it the ability to file, classify, and manage bugs without human intervention.
Prerequisites
Before your AI agent can start filing bugs, you'll need a human to complete a one-time setup:
A human needs to create a bugAgent account. After the workspace setup, bugAgent takes the new owner directly to API keys.
From Settings → Developers → API Keys, generate an API key. This is what your agent uses to authenticate. The full key is shown only once — save it securely.
Set the key as an environment variable or pass it via config. Your agent can now create bug reports, query reports, manage projects, and more.
Free Tier
Every bugAgent account starts on the Free plan. Your agent gets access to:
external_run_id resumes the existing run and does not consume another run. Deleting data does not reset monthly run usage. Browser, model, and network costs remain customer-side; restrict target access and network egress.The 128 KB Free structured-content bound is separate from file attachments. AI test-case generation, AI tag suggestions, Figma import, and test-case file attachments require Enterprise. Free can use URL references. Enterprise test case storage and runs are unlimited, subject to general platform protections.
Jira Cloud sync is included on Free. View the Enterprise plan for higher limits, advanced testing, team features, and priority support.
bugAgent MCP Server
The hosted bugAgent MCP server connects your AI agent to bugAgent via the Model Context Protocol. It uses Streamable HTTP and gives your agent direct access to:
Create, list, get, update, and classify bug reports. Auto-classification enriches every report with type, severity, and confidence score.
Create and manage projects. File bugs into specific projects for organized tracking across multiple codebases.
Check usage limits, view statistics, and get breakdowns by type, severity, and status. Your agent can monitor its own quota.
Workspace API keys are created and rotated by a human in Developer settings. Delegated OAuth sessions can use interactive API-key administration tools when authorized.
Sync bug reports to Jira with mapped fields, priority, and labels. Bridge AI-discovered bugs directly into your team's workflow.
Register, login, update profile, change password, and manage notification preferences.
Setup
Recommended: hosted Streamable HTTP
Connect directly to https://mcp.bugagent.com/mcp with the header Authorization: Bearer ba_live_YOUR_KEY_HERE. Nothing needs to run locally. Here are examples for popular clients:
claude mcp add --transport http bugagent https://mcp.bugagent.com/mcp \
--header "Authorization: Bearer ba_live_YOUR_KEY_HERE" // .cursor/mcp.json
{
"mcpServers": {
"bugagent": {
"type": "http",
"url": "https://mcp.bugagent.com/mcp",
"headers": {
"Authorization": "Bearer ba_live_YOUR_KEY_HERE"
}
}
}
} # Add to ~/.codex/config.toml
[mcp_servers.bugagent]
url = "https://mcp.bugagent.com/mcp"
bearer_token_env_var = "BUGAGENT_API_KEY" Export the API key before starting or restarting Codex:
export BUGAGENT_API_KEY="ba_live_YOUR_KEY_HERE" {
"mcpServers": {
"bugagent": {
"type": "http",
"url": "https://mcp.bugagent.com/mcp",
"headers": {
"Authorization": "Bearer ba_live_YOUR_KEY_HERE"
}
}
}
} Replace ba_live_YOUR_KEY_HERE with your actual API key from Settings → Developers.
Optional stdio bridge
Use the published bridge only when a client requires stdio and cannot connect to remote Streamable HTTP. The bridge command is npx -y bugagent-mcp:
{
"mcpServers": {
"bugagent": {
"command": "npx",
"args": ["-y", "bugagent-mcp"],
"env": { "BUGAGENT_API_KEY": "ba_live_YOUR_KEY_HERE" }
}
}
} Available MCP Tools
The complete delegated OAuth catalog contains 129 tools. This table is a short onboarding sample, not the canonical catalog. API-key clients see only tools allowed by their selected scopes; account, team, and API-key administration entries require OAuth.
Use the canonical MCP guide for human-readable details or download the generated mcp-tool-index.json for all runtime tools and access metadata.
create_bug_report File a new bug report. Auto-classifies bug type if not provided. list_bug_reports List bug reports with optional filters by type, severity, or search query. get_bug_report Get full details of a specific bug report by ID. update_bug_report Update an existing bug report. Only provided fields will be changed. classify_bug Classify a description into a bug type with confidence score. flush_reports Bulk delete old bug reports. Admin only. get_usage Check current usage against plan limits and remaining quota. get_stats Get daily counts, breakdowns by type, severity, and status. list_projects List available projects with names, slugs, and default status. create_project Create a new project. First project becomes the default. delete_project Delete a project. Choose how to handle its bug reports. register_account Create a new account with email and password. login Sign in and receive access tokens. update_profile Update your display name. change_password Change your account password. get_settings Get profile info, plan, and notification preferences. update_settings Update notification preferences. generate_api_key Create a named API key. Full key shown only once. list_api_keys List active keys. Shows prefix only, never the full key. regenerate_api_key Revoke and replace a key with the same name and scopes. delete_api_key Revoke and delete a key. Stops working immediately. sync_to_jira Sync a report to Jira with mapped fields, priority, and labels. upgrade_plan Request paid-plan enrollment. Returns the sales contact page. list_automations List Playwright automation scripts. Filter by project_id or status (draft/active/paused). get_automation Get full automation details including Playwright script and recent runs. run_automation Trigger an immediate Playwright test run in headless Chromium. Returns run_id and status. list_automation_runs List recent runs for an automation with status, duration, and errors. Quick Start for Agents
Create a bug report programmatically using the REST API:
import os
import requests
api_key = os.environ["BUGAGENT_API_KEY"]
base_url = os.getenv("BUGAGENT_BASE_URL", "https://app.bugagent.com")
headers = {"Authorization": f"Bearer {api_key}"}
response = requests.get(
f"{base_url}/api/reports",
headers=headers,
params={"project_id": os.environ["BUGAGENT_PROJECT_ID"], "limit": 5},
timeout=20,
)
response.raise_for_status()
for report in response.json().get("reports", []):
print(report["short_id"], report["title"]) const apiKey = process.env.BUGAGENT_API_KEY;
const projectId = process.env.BUGAGENT_PROJECT_ID;
const baseUrl = process.env.BUGAGENT_BASE_URL ?? "https://app.bugagent.com";
if (!apiKey || !projectId) throw new Error("Set BUGAGENT_API_KEY and BUGAGENT_PROJECT_ID");
const url = new URL("/api/reports", baseUrl);
url.searchParams.set("project_id", projectId);
url.searchParams.set("limit", "5");
const response = await fetch(url, {
headers: { Authorization: `Bearer ${apiKey}` },
signal: AbortSignal.timeout(20_000),
});
if (!response.ok) throw new Error(`bugAgent returned ${response.status}: ${await response.text()}`);
const payload = await response.json();
for (const report of payload.reports ?? []) console.log(report.short_id, report.title); What Makes bugAgent Different?
Unlike traditional bug tracking platforms, bugAgent is built from the ground up for AI-native workflows.
- AI auto-classifies bugs by type and severity
- MCP-native — agents file bugs directly
- Natural language input — no forms or templates
- Confidence scoring on every classification
- API-first with MCP, REST, and CLI
- Enrichment pipeline adds context automatically
- Zero context-switching for developers using AI
- Free tier for agents to get started instantly
- Manual classification by humans
- GUI-first — agents can't easily interact
- Rigid forms with required fields
- No intelligence on reports
- REST API bolted on as afterthought
- Requires manual triage and tagging
- Context-switching between IDE and browser
- Per-seat pricing, complex onboarding
Next Steps
Ready to dive deeper? Explore these resources: