# AI use cases for Carrel (FengZhenfei/carrel) Repo: [FengZhenfei/carrel](https://github.com/FengZhenfei/carrel) | Use cases: 4 | Page: https://dedupedai.com/use-cases/fengzhenfei-carrel/ Carrel is a self-hosted knowledge base that parses, indexes, and searches your documents so AI agents can retrieve source passages, images, and related facts through an API. This file holds every use case of this repo. Each one starts with a "## " heading and ends with its URL. - Let a coding agent search your product docs: https://dedupedai.com/use-cases/fengzhenfei-carrel/let-coding-agent-search-product-docs/ - Answer contract questions with cited passages: https://dedupedai.com/use-cases/fengzhenfei-carrel/answer-contract-questions-with-citations/ - Trace decisions and conflicts across meeting notes: https://dedupedai.com/use-cases/fengzhenfei-carrel/trace-decisions-across-meeting-notes/ - Keep a research library current as files change: https://dedupedai.com/use-cases/fengzhenfei-carrel/keep-research-library-current-automatically/ ## How to let a coding agent search your product docs with Carrel Repo: [FengZhenfei/carrel](https://github.com/FengZhenfei/carrel) | Difficulty: Intermediate | Time: 90 minutes | Basis: From the README Deploy Carrel locally and give Claude Code or Codex a search skill over your own document folders. **Who it is for:** A developer or small team whose coding agent keeps guessing at product specs, internal guides, or old design notes stored in local files. **The problem:** The agent has no access to the team's documents, so answers are generic or wrong. **The outcome:** The agent answers questions from the actual documents and can cite the source passages it used. ### Steps 1. Run `git clone https://github.com/FengZhenfei/carrel.git` and then `cd carrel`. 2. Run `./deploy.sh` to start the services and create the configuration files. 3. Edit `config/knowledge-base.env` to set your embedding and vision model URLs, model names, and API keys. 4. Put your documents in `runtime/mirror//`, open http://127.0.0.1:9800, and enable that folder. 5. Install the bundled carrel-search skill in your agent, then confirm a query with `curl -sS http://127.0.0.1:9810/search -H 'Content-Type: application/json' -d '{"question": "What are the delivery and acceptance requirements?", "top_k": 8}'`. (inferred, check the README) ### Prompt to paste into your agent ```text Using the Carrel repo at https://github.com/FengZhenfei/carrel, set up the carrel-search skill for this agent and query the search endpoint at http://127.0.0.1:9810/search with a question about . Return the answer with the source passages cited. Verify that the response contains source passages before answering. ``` ### Caveat Carrel sends documents to the model endpoints you configure, so hosted endpoints mean your files leave your machine unless you run local models. 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/fengzhenfei-carrel/let-coding-agent-search-product-docs/ ## How to answer contract questions with cited passages with Carrel Repo: [FengZhenfei/carrel](https://github.com/FengZhenfei/carrel) | Difficulty: Intermediate | Time: 90 minutes | Basis: From the README Index a folder of contracts and let an agent pull the clauses it needs before it answers. **Who it is for:** A legal, procurement, or operations person who needs quick answers about delivery, acceptance, or payment terms across many contract files. **The problem:** Finding the right clause across many PDFs and Word files is slow, and an agent without the files will invent terms. **The outcome:** The agent returns answers grounded in retrieved contract passages, with the source text available for review. ### Steps 1. Run `./deploy.sh` from the cloned Carrel folder and wait for the first startup to finish downloading parser models. 2. Copy the contract files into `runtime/mirror//` and enable that folder in the console at http://127.0.0.1:9800. 3. Check service health and task progress in the console before your first query. 4. Run `curl -sS http://127.0.0.1:9810/search -H 'Content-Type: application/json' -d '{"question": "What are the delivery and acceptance requirements?", "top_k": 8}'` to test retrieval. (inferred, check the README) ### Prompt to paste into your agent ```text Use the Carrel search API in the FengZhenfei/carrel repo at http://127.0.0.1:9810/search to find the clauses about across the enabled contract folder. Report each answer with the source passage it came from. Verify that every claim in the answer appears in a returned passage. ``` ### Caveat Carrel does not state that its answers are legally reliable, so a person should check any clause before relying on it. Risk, credentials: If you enable LAN access, set both KB_WEB_HOST and KB_WEB_TOKEN and set KB_SEARCH_TOKEN before using protected search endpoints remotely, since contracts are confidential. 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/fengzhenfei-carrel/answer-contract-questions-with-citations/ ## How to trace decisions and conflicts across meeting notes with Carrel Repo: [FengZhenfei/carrel](https://github.com/FengZhenfei/carrel) | Difficulty: Advanced | Time: 2 hours | Basis: From the README Build a knowledge graph over meeting notes to follow relationships, sources, and conflicting facts. **Who it is for:** A project lead or analyst who needs to see how a decision changed across many weeks of meeting notes. **The problem:** Plain keyword or vector search finds passages but does not show how entities, decisions, and dates connect or where facts conflict. **The outcome:** The agent can follow graph relationships and report facts with their sources, time information, and conflict indicators. ### Steps 1. Run `./deploy.sh` and set your embedding and vision endpoints in `config/knowledge-base.env`. 2. Register a chat model in the console at http://127.0.0.1:9800 and select it in the knowledge-base settings. 3. Enable graph building for the meeting-notes folder under `runtime/mirror//`. 4. Wait for the scheduled tasks to finish, then query the search endpoint with `curl -sS http://127.0.0.1:9810/search -H 'Content-Type: application/json' -d '{"question": "Which decisions about changed over time?", "top_k": 8}'`. (inferred, check the README) ### Prompt to paste into your agent ```text Using the FengZhenfei/carrel search API at http://127.0.0.1:9810/search, ask about how changed across the meeting notes, and follow graph relationships when more evidence is needed. Return a dated timeline with sources and mark any conflicting facts. Verify that each dated item links to a source passage. ``` ### Caveat Knowledge graph construction needs a chat model that you configure, and the README does not describe how accurate the extracted facts are. 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/fengzhenfei-carrel/trace-decisions-across-meeting-notes/ ## How to keep a research library current as files change with Carrel Repo: [FengZhenfei/carrel](https://github.com/FengZhenfei/carrel) | Difficulty: Intermediate | Time: 90 minutes | Basis: From the README Schedule Carrel to scan a research folder so new papers and notes show up in agent answers. **Who it is for:** A researcher or analyst whose reading list and notes change every week and who wants agents to use the latest versions. **The problem:** Indexes go stale when files are added or edited, so agents answer from outdated material. **The outcome:** Carrel scans for file changes and updates indexes on a schedule, so agent queries reflect recent files. ### Steps 1. Run `./deploy.sh` and wait for the services to start; on Linux with user systemd, the script enables scheduled tasks. 2. Place the research files in `runtime/mirror//` and enable that folder in the console at http://127.0.0.1:9800. 3. On Linux without user systemd, follow the systemd guide to set up the timers manually, using deployment/systemd/README.md. (inferred, check the README) 4. Check task progress in the console after you add or change a file, then rerun the search query to confirm the change appears. ### Prompt to paste into your agent ```text Use the FengZhenfei/carrel repo to search the research library at http://127.0.0.1:9810/search for . Report whether results include a document added in the last week. Verify the returned sources are dated or named after the new file before reporting success. ``` ### Caveat Scheduled updates depend on the host setup, and the README says macOS needs scheduling arranged separately while Windows requires WSL2. 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/fengzhenfei-carrel/keep-research-library-current-automatically/