# AI use cases for System One Connector (itsmostafa/system-one-connector) Repo: [itsmostafa/system-one-connector](https://github.com/itsmostafa/system-one-connector) | Use cases: 4 | Page: https://dedupedai.com/use-cases/itsmostafa-system-one-connector/ A Go MCP connector that gives AI agents direct access to System One models, mainly TypeSafe's Jev, for quick yes/no, choice and score judgments with probabilities. This file holds every use case of this repo. Each one starts with a "## " heading and ends with its URL. - Triage support tickets by urgency and owning team: https://dedupedai.com/use-cases/itsmostafa-system-one-connector/triage-tickets-by-urgency-and-team/ - Classify a dataset of records in one call: https://dedupedai.com/use-cases/itsmostafa-system-one-connector/batch-classify-records-with-items/ - Let an agent act only when it is confident: https://dedupedai.com/use-cases/itsmostafa-system-one-connector/gate-agent-actions-on-confidence/ - Route judgments to a self-hosted CLM or Laya model: https://dedupedai.com/use-cases/itsmostafa-system-one-connector/run-self-hosted-clm-or-laya-model/ ## How to triage support tickets by urgency and owning team with System One Connector Repo: [itsmostafa/system-one-connector](https://github.com/itsmostafa/system-one-connector) | Difficulty: Intermediate | Time: 15 minutes | Basis: From the README Have an agent score each ticket for urgency and pick the owning team with probabilities. **Who it is for:** A support lead or support engineer whose agent receives incoming tickets and needs to route them without reading each one. **The problem:** An LLM reply like 'this seems fairly urgent' gives an agent nothing to branch on, and it cannot tell a confident answer from a guess. **The outcome:** Each ticket gets a yes/no urgency probability and a team choice with per-team probabilities that the agent can compare against a threshold. ### Steps 1. Run `curl -fsSL https://raw.githubusercontent.com/itsmostafa/system-one-connector/main/install.sh | sh` 2. Get a key at https://console.typesafe.ai/ and run `TYPESAFE_API_KEY=your-key evaluate setup mcp` 3. Open Claude Code or Codex and ask it to use evaluate on a ticket with an urgency question and a department question 4. Read the returned `is_urgent` noul value and the `department` probabilities 5. Route the ticket when the top probability is high, and send unsure tickets to a person ### Prompt to paste into your agent ```text Use the evaluate tool from the itsmostafa/system-one-connector repo to decide whether this ticket is urgent (noul question) and which team should own it (choice question with billing, technical and sales criteria): . Return the noul value and the department probabilities as JSON. Verify that the probabilities for the choice question sum to about 1. ``` ### Caveat The README's example teams (billing, technical, sales) are placeholders, so you must define your own criteria for your team structure. 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/itsmostafa-system-one-connector/triage-tickets-by-urgency-and-team/ ## How to classify a dataset of records in one call with System One Connector Repo: [itsmostafa/system-one-connector](https://github.com/itsmostafa/system-one-connector) | Difficulty: Intermediate | Time: 30 minutes | Basis: From the README Pass up to 500 records as items and ask the same questions of each one. **Who it is for:** A data analyst or ops engineer with a spreadsheet or export of customer messages, feedback or logs to label. **The problem:** Labeling rows one by one through a chat model is slow, inconsistent and hard to check at scale. **The outcome:** Each record returns its own answers, and a failure on one record does not stop the rest from completing. ### Steps 1. Run `curl -fsSL https://raw.githubusercontent.com/itsmostafa/system-one-connector/main/install.sh | sh` 2. Run `TYPESAFE_API_KEY=your-key evaluate setup mcp` with your key 3. Export your records to a JSON array, with up to 500 entries (inferred, check the README) 4. Ask your agent to call evaluate with the records as `items` and the same questions for all of them; see docs/tool-reference.md for the exact input fields 5. Check the per-record results and re-run only the records that failed ### Prompt to paste into your agent ```text Using the evaluate tool from the itsmostafa/system-one-connector repo, pass the records in as items (up to 500) and ask whether each one is negative in sentiment using a score question. Return one answer per record as CSV. Verify that the number of returned rows equals the number of input records, and list any that failed. ``` ### Caveat The README says up to 500 records per call and points to docs/tool-reference.md for the exact item format, so check that file before building the input. 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/itsmostafa-system-one-connector/batch-classify-records-with-items/ ## How to let an agent act only when it is confident with System One Connector Repo: [itsmostafa/system-one-connector](https://github.com/itsmostafa/system-one-connector) | Difficulty: Intermediate | Time: 30 minutes | Basis: From the README Use Jev confidence scores to proceed on clear cases and escalate unsure ones. **Who it is for:** A developer building an agent workflow that takes actions such as replies, refunds or deletions and needs a safe stop point. **The problem:** An agent that cannot measure its own certainty either acts on guesses or asks a human about everything. **The outcome:** The agent proceeds when the confidence is high and escalates to a person when the answer is close to a coin flip. ### Steps 1. Run `curl -fsSL https://raw.githubusercontent.com/itsmostafa/system-one-connector/main/install.sh | sh` 2. Run `TYPESAFE_API_KEY=your-key evaluate setup mcp` 3. Write a rule in your agent's instructions: proceed when the top probability is above your chosen threshold, otherwise ask a person 4. Test the rule on a few known cases from your own history 5. Adjust the threshold until the escalations match what a person would have escalated ### Prompt to paste into your agent ```text Use the evaluate tool from the itsmostafa/system-one-connector repo to answer this question about : . Proceed only if the top probability is at least , otherwise return 'escalate'. Verify the returned confidence field matches the highest probability before deciding. ``` ### Caveat The threshold is your choice and the README gives no recommended value, so test it on your own cases before relying on it. 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/itsmostafa-system-one-connector/gate-agent-actions-on-confidence/ ## How to route judgments to a self-hosted CLM or Laya model with System One Connector Repo: [itsmostafa/system-one-connector](https://github.com/itsmostafa/system-one-connector) | Difficulty: Advanced | Time: 1 hour | Basis: From the README Point the connector at your own server to run CLM or Laya instead of the hosted Jev. **Who it is for:** A developer or team that wants judgment calls to run on a model server they control. **The problem:** Some teams cannot send their data to a hosted API and need the same question-and-probability workflow on their own infrastructure. **The outcome:** Your agent gets the same typed answers from a self-hosted model, with the connector configured to use your own endpoint. ### Steps 1. Run `curl -fsSL https://raw.githubusercontent.com/itsmostafa/system-one-connector/main/install.sh | sh` 2. Start your own model server and note its address (the README shows an example at http://127.0.0.1:8787) 3. For Laya, run `TYPESAFE_API_KEY=local TYPESAFE_BASE_URL=http://127.0.0.1:8787 evaluate setup mcp` 4. For CLM, also set `TYPESAFE_MODEL=clm-latest`; see docs/configuration.md#running-clm-locally 5. Save each setup as a profile and switch with `evaluate profile use ` (inferred, check the README) ### Prompt to paste into your agent ```text Using the evaluate tool from the itsmostafa/system-one-connector repo configured with a self-hosted CLM or Laya endpoint, ask whether conveys urgency (noul question). Return the probability and the endpoint used. Verify that the response came from your local address and not a hosted one. ``` ### Caveat The README does not explain how to run the model server itself, so you need your own working CLM or Laya server before this setup works. 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/itsmostafa-system-one-connector/run-self-hosted-clm-or-laya-model/