> ## Documentation Index
> Fetch the complete documentation index at: https://docs.firedrill.run/llms.txt
> Use this file to discover all available pages before exploring further.

# AI agent examples by framework and stack

> Find AI agent simulation examples for LangGraph, Pydantic AI, Google ADK, Mastra, CopilotKit, browser automation, background workers, and multi-agent teams.

Choose an example with the framework and entry point closest to your application. These 40 projects cover Python, JavaScript, TypeScript, browser agents, web copilots, background workers, and multi-agent teams.

> [!NOTE]
>
> Repository availability does not grant a license to reuse the code. Follow each
> repository's setup instructions and check its license terms before reuse.

## Start with your own agent

Run this in your application's repository, or give it to your coding agent:

```sh theme={null}
npx firedrill-run init
```

Review the optional setup choices, authenticate in the browser, and select your project. Then connect your agent's [real entry point](/guides/example-agents). The [quickstart](/quickstart) explains the complete flow; [native MCP setup](/guides/coding-agents) connects your coding agent separately.

The examples use isolated local synthetic fixtures and observation-only Firedrill recipes. Their adapters invoke the application's actual entry point. A local fixture is not a managed Firedrill Tool instance, and an observation-only recipe does not declare a quality verdict. Follow each repository's setup instructions for its runtime and model-provider requirements.

## Python agent frameworks

These command-based agents cover structured outputs, cited answers, stateful operations, and configuration comparisons.

| Framework and example | Entry point | What to explore |
| - | - | - |
| [Pydantic AI support ticket routing](https://github.com/firedrill-tools/support-ticket-routing-agent) | Python CLI | Ambiguous requests, clarification, and out-of-scope tickets. |
| [LangChain company policy answers](https://github.com/firedrill-tools/company-policy-answer-agent) | Python CLI and local tools | Source citations, effective dates, and conflicting policies. |
| [Gemini invoice and receipt extraction](https://github.com/firedrill-tools/invoice-receipt-extraction-agent) | Python CLI | Missing fields, unreadable values, and structured extraction. |
| [LlamaIndex research briefs](https://github.com/firedrill-tools/research-brief-agent) | Python CLI and local sources | Source maps and contradictory claims. |
| [OpenAI Agents SDK meeting scheduling](https://github.com/firedrill-tools/meeting-scheduling-agent) | Python CLI and local tools | Time zones, changing availability, and booking revisions. |
| [LangChain sales qualification with Composio](https://github.com/firedrill-tools/sales-qualification-agent) | Python CLI | Synthetic CRM, email, and calendar actions; separate connector configuration. |
| [LangGraph Revenue Desk](https://github.com/firedrill-tools/revenue-desk-agent) | Python CLI and local tools | Cross-system revenue requests, recovery, and synthetic state changes. |
| [Pydantic AI refund resolution](https://github.com/firedrill-tools/refund-resolution-agent) | Python CLI and local tools | Current identity, approval, currency, and refundable balance. |
| [Google ADK accounts payable](https://github.com/firedrill-tools/accounts-payable-agent) | Python CLI and local ledger | Invoice matching, approval requests, and ledger changes. |
| [CrewAI purchase requests](https://github.com/firedrill-tools/purchase-request-agent) | Python CLI and local tools | Quotes, stock, budget, and named approval before ordering. |
| [OpenAI Agents SDK infrastructure operations](https://github.com/firedrill-tools/infrastructure-operations-agent) | Python and a synthetic CLI | Deployment status, failed rollouts, health checks, and rollback. |
| [Pydantic AI order-exception model comparison](https://github.com/firedrill-tools/order-exception-model-comparison) | Python CLI benchmark | The same order cases across configured OpenAI, Anthropic, and Gemini providers. |

## JavaScript and TypeScript agents

Use these examples for conversational workflows, MCP-backed business actions, and agents that work with files or repositories.

| Framework and example | Stack and interface | What to explore |
| - | - | - |
| [Vercel AI SDK onboarding adviser](https://github.com/firedrill-tools/onboarding-adviser-agent) | JavaScript CLI conversation | Corrections, retractions, and requirements gathered over multiple turns. |
| [Claude Agent SDK inbox operations](https://github.com/firedrill-tools/claude-inbox-operations-agent) | TypeScript; local or HTTP MCP | Synthetic mailbox search, labels, sending permissions, and retries. |
| [Mastra project coordination](https://github.com/firedrill-tools/project-coordination-agent) | TypeScript CLI and local tools | Exact issue identity, owner channels, and completion evidence. |
| [Mastra collections follow-up](https://github.com/firedrill-tools/collections-followup-agent) | TypeScript CLI and local tools | Payments, promises, and CRM state at dispatch. |
| [Codex SDK repository repair](https://github.com/firedrill-tools/codex-repository-repair-agent) | JavaScript; filesystem and CLI | Repairs in a disposable checkout with a narrow changed-file boundary. |
| [Codex SDK dependency maintenance](https://github.com/firedrill-tools/codex-dependency-maintenance-agent) | JavaScript; filesystem and CLI | Dependency changes, lockfiles, and application behavior. |
| [Claude Agent SDK data repair](https://github.com/firedrill-tools/data-repair-agent) | JavaScript and filesystem MCP | Synthetic CSV and JSON repairs without changing unrelated records. |
| [Claude Agent SDK release preparation](https://github.com/firedrill-tools/release-preparation-agent) | JavaScript; Git CLI and MCP | Local release drafts, commit ranges, and publication safeguards. |

## Web apps and AG-UI copilots

These examples keep the user interface and backend agent connected. Explicit user decisions remain separate from model proposals.

The included Firedrill adapters exercise the backend HTTP or AG-UI seam, not the frontend. Add a [browser workflow](/guides/browser-testing) when you need evidence of user-interface behavior.

| Framework and example | Stack and interface | What to explore |
| - | - | - |
| [CopilotKit and LangGraph support workspace](https://github.com/firedrill-tools/support-copilot-workspace) | Python + TypeScript; HTTP / AG-UI | Case inspection, refund proposals, and operator decisions. |
| [CopilotKit and LangChain document review](https://github.com/firedrill-tools/document-review-workspace) | Python + TypeScript; HTTP / AG-UI | Section-level suggestions and explicitly approved edits. |
| [Vercel AI SDK shopping assistant](https://github.com/firedrill-tools/shopping-assistant-app) | TypeScript web app and local SQLite | Read-only advice, cart budgets, stock, and checkout. |
| [CopilotKit and Google ADK onboarding](https://github.com/firedrill-tools/onboarding-copilot-workspace) | Python + TypeScript; HTTP / AG-UI | Invitation proposals and committed team membership. |
| [CopilotKit and Mastra incident coordination](https://github.com/firedrill-tools/incident-coordination-workspace) | TypeScript; HTTP / AG-UI | Telemetry, owner approvals, and synthetic channel updates. |

## Browser agents and Playwright

These JavaScript applications act through a browser rather than replacing the workflow with direct database writes. Browser evidence depends on the example's capture configuration; not every API run has a screenshot.

| Framework and example | Browser interface | What to explore |
| - | - | - |
| [OpenAI Responses admin portal operator](https://github.com/firedrill-tools/admin-portal-operator) | Playwright | Directory operations through browser actions and screenshot traces. |
| [Anthropic invoice portal processor](https://github.com/firedrill-tools/invoice-portal-processor) | Playwright | Approved invoices, exact totals, missing fields, and duplicates. |
| [Gemini booking browser assistant](https://github.com/firedrill-tools/booking-browser-assistant) | Playwright | Availability review and time-zone-aware appointment confirmation. |
| [Claude Agent SDK CRM browser operator](https://github.com/firedrill-tools/crm-browser-operator) | Browser MCP | Contact identity, duplicate records, and stale revisions. |
| [Vercel AI SDK support console operator](https://github.com/firedrill-tools/support-console-operator) | Playwright | Attachments, later corrections, access boundaries, and failed submits. |

## Event-driven and background agents

Use explicit events, jobs, or scheduled ticks to explore work that extends beyond one request. Local application clocks and journals are not automatically the managed virtual clock.

| Framework and example | Stack and interface | What to explore |
| - | - | - |
| [Gemini lead intake with Pipedream](https://github.com/firedrill-tools/lead-intake-agent) | Python webhook | Inbound form routing and acknowledged synthetic connector actions. |
| [LangGraph webhook incident responder](https://github.com/firedrill-tools/webhook-incident-responder) | Python; HTTP webhook / Slack seam | Ordered events, retained incident state, and notification queues. |
| [Google ADK async document processing](https://github.com/firedrill-tools/async-document-processing-agent) | Python background worker | Correlated jobs, per-attempt evidence, recovery, and ambiguity review. |
| [LangGraph renewal management](https://github.com/firedrill-tools/renewal-management-agent) | Python scheduled workflow | Payment changes, decisions, deadlines, and follow-ups at each tick. |
| [Mastra inventory monitoring](https://github.com/firedrill-tools/inventory-monitoring-agent) | TypeScript event worker | Ordered stock changes, low-stock alerts, and replenishment drafts. |
| [Anthropic support escalation with Pipedream](https://github.com/firedrill-tools/support-escalation-worker) | Python event worker | Clock events, late replies, resolved tickets, and duplicate escalation. |

## Multi-agent teams

These Python projects explore specialist coordination, shared state, handoffs, and disagreement rather than treating every agent call as an independent answer.

| Framework and example | Coordination interface | What to explore |
| - | - | - |
| [CrewAI research team](https://github.com/firedrill-tools/research-team-agent) | Local sources and role-based CLI team | Research, challenge, writing, citations, and conflicting sources. |
| [OpenAI Agents SDK customer escalation team](https://github.com/firedrill-tools/customer-escalation-team) | Specialist handoffs | Triage, billing, account specialists, and current permissions. |
| [LangGraph contract review team](https://github.com/firedrill-tools/contract-review-team) | Local contract workflow | Sourced terms, risk reviews, conflicting notices, and reviewer disagreement. |
| [Google ADK resource allocation team](https://github.com/firedrill-tools/resource-allocation-team) | Shared SQLite resources | Competing room/equipment holds, transaction conflicts, and compensation. |

## Turn an example into your simulation

1. Start with one small scenario and the application's real entry point. Keep its model credentials in your own runner.
2. Review the recipe's inputs, selected configuration, and isolation strategy. Local synthetic dependencies do not authorize actions on real accounts.
3. Run an observation-only simulation and inspect the [recorded output and evidence](/guides/results). A completed execution is not automatically a pass.
4. Add [quality scoring](/guides/scoring) for acceptable output variation, or exact checks for hard constraints, when they answer your question.
5. Reuse the reviewed recipe in [CI](/guides/ci), add [scenario variations](/guides/scenario-matrices), or compare [model configurations](/guides/model-benchmarks).

Source examples alone do not prove live model quality, real connector delivery, hosted Tool availability, or saved cloud reports. Those depend on the accounts, configuration, and actual execution you authorize.


This documentation is built and hosted on [Mintlify](https://mintlify.com), a developer documentation platform.