# AI use cases for Work Wit AI Agent (crane-in-clear-sky/WorkWit-AI-Agent)

Repo: [crane-in-clear-sky/WorkWit-AI-Agent](https://github.com/crane-in-clear-sky/WorkWit-AI-Agent) | Use cases: 4 | Page: https://dedupedai.com/use-cases/crane-in-clear-sky-workwit-ai-agent/

WorkWit is a self-hosted, MIT-licensed AI Agent platform for office work, with built-in contract review and resume screening, deployed with one Docker Compose command.

This file holds every use case of this repo. Each one starts with a "## " heading and ends with its URL.

- Deploy a self-hosted office AI agent on your own server: https://dedupedai.com/use-cases/crane-in-clear-sky-workwit-ai-agent/deploy-self-hosted-office-agent/
- Score resumes against a job profile in minutes: https://dedupedai.com/use-cases/crane-in-clear-sky-workwit-ai-agent/screen-resumes-against-job-profile/
- Get contract risk points from your side's stance: https://dedupedai.com/use-cases/crane-in-clear-sky-workwit-ai-agent/review-contract-risks-by-stance/
- Add an MCP server's tools to the WorkWit agent: https://dedupedai.com/use-cases/crane-in-clear-sky-workwit-ai-agent/connect-mcp-server-to-office-agent/

## How to deploy a self-hosted office AI agent on your own server with Work Wit AI Agent

Repo: [crane-in-clear-sky/WorkWit-AI-Agent](https://github.com/crane-in-clear-sky/WorkWit-AI-Agent) | Difficulty: Intermediate | Time: 2 hours | Basis: From the README

Run WorkWit on a Linux server with Docker Compose and connect it to your own model endpoints.

**Who it is for:** An IT admin at a small organization who wants AI office tools without a per-seat SaaS subscription or sending company files to a vendor.

**The problem:** Office AI tools are usually hosted by a vendor, which means company contracts and HR files leave the organization's infrastructure.

**The outcome:** A running WorkWit instance on port 4001 that uses your own model services, with the default admin password changed.

### Steps

1. Clone the repo: `git clone https://github.com/crane-in-clear-sky/WorkWit-AI-Agent ~/workwit`, then run `cd ~/workwit` (inferred, check the README)
2. Install Docker if needed: `curl -fsSL https://get.docker.com | sudo sh`, then `sudo usermod -aG docker $USER`
3. Copy the template: `cp .env.example .env`, then set the four MODEL_*_BASE and MODEL_*_KEY pairs to your real addresses and keys
4. Start the stack: `docker compose up -d --build`, then check `docker compose logs -f backend` for errors
5. Open http://<server-address>:4001, sign in as admin, and change the default password under System Management -> User Management

### Prompt to paste into your agent

```text
In the WorkWit-AI-Agent repo at <repo-path>, set up .env with the four model endpoints at <model-base-url> and <model-key>, then start the stack with docker compose up -d --build. Expected output: the backend container running and the web app reachable on port 4001. Verify that docker compose logs shows no startup errors and that the admin password has been changed.
```

### Caveat

The README's git clone line is incomplete in the source, so the clone URL in step 1 is inferred and should be checked against the repo page.

Disclaimer: Third-party repos we don't own or vouch for. AI-generated use cases can be wrong: review everything and run at your own risk. Full text: https://dedupedai.com/disclaimer/

URL: https://dedupedai.com/use-cases/crane-in-clear-sky-workwit-ai-agent/deploy-self-hosted-office-agent/

## How to score resumes against a job profile in minutes with Work Wit AI Agent

Repo: [crane-in-clear-sky/WorkWit-AI-Agent](https://github.com/crane-in-clear-sky/WorkWit-AI-Agent) | Difficulty: Intermediate | Time: 2 hours | Basis: From the README

Upload a resume and a job profile to get key candidate fields, a match score, and a recommendation.

**Who it is for:** An HR recruiter with a batch of applicants for one open role who needs a first-pass ranking.

**The problem:** Reading every resume against a job description by hand takes time, and the comparison is inconsistent across reviewers.

**The outcome:** For each resume, a list of key fields, a job-profile match score, and a recommendation to guide the next review step.

### Steps

1. Deploy WorkWit and sign in to the web app on port 4001
2. Open the Resume Screening feature
3. Upload one candidate resume and the job profile for the open role
4. Read the extracted key fields, the match score, and the recommendation
5. Repeat for each candidate and review the top scores yourself before any decision

### Prompt to paste into your agent

```text
Using the WorkWit-AI-Agent repo deployed at <app-url>, run Resume Screening on <resume-file> against the job profile <job-profile-file>. Expected output: key candidate fields, a match score, and a recommendation. Verify that the score is explained by fields that appear in the resume text.
```

### Caveat

The README describes the match score and recommendation as model output and does not describe any validation, so a human should check them before acting on a candidate.

Disclaimer: Third-party repos we don't own or vouch for. AI-generated use cases can be wrong: review everything and run at your own risk. Full text: https://dedupedai.com/disclaimer/

URL: https://dedupedai.com/use-cases/crane-in-clear-sky-workwit-ai-agent/screen-resumes-against-job-profile/

## How to get contract risk points from your side's stance with Work Wit AI Agent

Repo: [crane-in-clear-sky/WorkWit-AI-Agent](https://github.com/crane-in-clear-sky/WorkWit-AI-Agent) | Difficulty: Beginner | Time: 1 hour | Basis: From the README

Upload a contract as txt, docx, or pdf and get risk points and revision suggestions for Party A or Party B.

**Who it is for:** A business owner or ops lead who has to sign a vendor or partner contract and wants a first read of the risks for their side.

**The problem:** Reading a long contract for risky clauses takes time, and it is easy to miss clauses that matter for one side.

**The outcome:** A structured list of risk points and suggested revisions written from the stance you chose.

### Steps

1. Deploy WorkWit and sign in to the web app on port 4001
2. Open the Contract Review feature
3. Upload the contract as txt, docx, or pdf
4. Select your stance, Party A or Party B
5. Read the risk points and revision suggestions, then take the flagged clauses to your legal reviewer

### Prompt to paste into your agent

```text
Using the WorkWit-AI-Agent repo at <app-url>, run Contract Review on <contract-file> with the stance set to <Party A or Party B>. Expected output: structured risk points and revision suggestions. Verify that each risk points to a clause that actually appears in the uploaded contract.
```

### Caveat

The README notes a truncated-JSON repair fallback when parsing fails, so a result may be incomplete and should be checked against the source document.

Disclaimer: Third-party repos we don't own or vouch for. AI-generated use cases can be wrong: review everything and run at your own risk. Full text: https://dedupedai.com/disclaimer/

URL: https://dedupedai.com/use-cases/crane-in-clear-sky-workwit-ai-agent/review-contract-risks-by-stance/

## How to add an MCP server's tools to the WorkWit agent with Work Wit AI Agent

Repo: [crane-in-clear-sky/WorkWit-AI-Agent](https://github.com/crane-in-clear-sky/WorkWit-AI-Agent) | Difficulty: Advanced | Time: 2 hours | Basis: From the README

Point WorkWit at an MCP server through data/mcp.json so its tools join the agent's catalog.

**Who it is for:** A developer or IT admin who already runs an MCP server with tools the office team needs, such as an internal system, and wants the agent to use them.

**The problem:** The agent can only use the tools it was given, so internal systems stay out of reach unless they are connected.

**The outcome:** External tools from your MCP server appear in the unified capability catalog and can be called by the agent.

### Steps

1. Deploy WorkWit and confirm the stack runs with `docker compose up -d --build`
2. Create data/mcp.json in the repo with the connector settings for your MCP server, using streamable-http or stdio transport (inferred, check the README)
3. Restart the backend with `docker compose up -d --build` so the connector loads (inferred, check the README)
4. Open the Tool Library and confirm the external tools are listed
5. Ask the agent to use one of those tools on a test task and check the result

### Prompt to paste into your agent

```text
In the WorkWit-AI-Agent repo at <repo-path>, add an MCP connector for <mcp-server-name> in data/mcp.json using the <transport> transport, then rebuild the stack. Expected output: the external tools listed in the Tool Library. Verify that a test agent task calls one of those tools and returns a result, and that data/mcp.json is not committed to git.
```

### Caveat

The README does not show the format of data/mcp.json, so the connector fields must be taken from your MCP server's own documentation and the repo's code.

Risk, credentials: MCP connector config may contain tokens, so keep data/mcp.json out of git and limit the tokens to the access the agent needs.

Disclaimer: Third-party repos we don't own or vouch for. AI-generated use cases can be wrong: review everything and run at your own risk. Full text: https://dedupedai.com/disclaimer/

URL: https://dedupedai.com/use-cases/crane-in-clear-sky-workwit-ai-agent/connect-mcp-server-to-office-agent/