# AI use cases for Poirot (HezaoHezao/poirot) Repo: [HezaoHezao/poirot](https://github.com/HezaoHezao/poirot) | Use cases: 4 | Page: https://dedupedai.com/use-cases/hezaohezao-poirot/ Poirot is an open-source deep research agent kernel in Python, built around a ReAct loop with long-term memory, skills, sandboxed execution, and multi-agent delegation. This file holds every use case of this repo. Each one starts with a "## " heading and ends with its URL. - Run a deep research session from the terminal: https://dedupedai.com/use-cases/hezaohezao-poirot/run-deep-research-session/ - Study the agent kernel and run its test suite: https://dedupedai.com/use-cases/hezaohezao-poirot/study-agent-kernel-with-tests/ - Turn on long-term memory and skill evolution for a recurring topic: https://dedupedai.com/use-cases/hezaohezao-poirot/enable-memory-and-skills-for-recurring-topic/ - Run research tasks inside a Docker sandbox on Windows: https://dedupedai.com/use-cases/hezaohezao-poirot/run-research-in-docker-sandbox/ ## How to run a deep research session from the terminal with Poirot Repo: [HezaoHezao/poirot](https://github.com/HezaoHezao/poirot) | Difficulty: Intermediate | Time: 30 minutes | Basis: From the README Install Poirot locally and ask it a research question in the full-screen TUI. **Who it is for:** A developer or analyst who needs a multi-step research answer and is willing to install a Python tool. **The problem:** Doing multi-step research by hand across many sources takes time, and a single chat prompt rarely covers it. **The outcome:** A running Poirot session that answers a research question in the TUI conversation view. ### Steps 1. Clone the repo from https://github.com/HezaoHezao/poirot and change into the project folder. (inferred, check the README) 2. Create a virtual environment with `python -m venv .venv`. 3. Activate it with `source .venv/bin/activate` on Linux or macOS, or `.venv\Scripts\activate` on Windows. 4. Install the project with `pip install -e ".[dev]"`. 5. Copy the sample config with `cp .env.example .env` and set DEEPSEEK_API_KEY in .env. 6. Start Poirot with `poirot` and type your research question. ### Prompt to paste into your agent ```text Using the Poirot repo at , create a virtual environment, install it, set DEEPSEEK_API_KEY in .env, and launch `poirot` with the research question . Expected output: a research answer shown in the TUI conversation view. Verify that the status bar shows live token usage. ``` ### Caveat The README says the default launch is the TUI and points to USAGE.md for full configuration and troubleshooting, so check that file if setup fails. 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/hezaohezao-poirot/run-deep-research-session/ ## How to study the agent kernel and run its test suite with Poirot Repo: [HezaoHezao/poirot](https://github.com/HezaoHezao/poirot) | Difficulty: Intermediate | Time: 30 minutes | Basis: From the README Install the dev dependencies, run the tests, and trace the middleware layers. **Who it is for:** An engineer or AI agent evaluating how a research agent is architected before adopting or forking its design. **The problem:** Architecture claims are hard to judge without reading the code and checking that the tests pass. **The outcome:** A checked-out copy of Poirot with its test suite run and a short map of the middleware and memory layers. ### Steps 1. Clone the repo from https://github.com/HezaoHezao/poirot and change into the project folder. (inferred, check the README) 2. Create a virtual environment with `python -m venv .venv`. 3. Activate it with `source .venv/bin/activate` on Linux or macOS, or `.venv\Scripts\activate` on Windows. 4. Install the dev extras with `pip install -e ".[dev]"`. 5. Run the test suite with `pytest`. (inferred, check the README) 6. Read the Core Modules and Architecture sections of the README, then list the middleware in the agents package. ### Prompt to paste into your agent ```text In the Poirot repo at , install the dev extras, run the test suite, and report the pass and fail counts. Expected output: a short summary of the test result and a list of the middleware named in the README. Verify that the README's claimed test count of 2400+ roughly matches the number of tests collected. ``` ### Caveat The README gives no test command, so the pytest step is inferred and may need adjusting to the project's test setup. 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/hezaohezao-poirot/study-agent-kernel-with-tests/ ## How to turn on long-term memory and skill evolution for a recurring topic with Poirot Repo: [HezaoHezao/poirot](https://github.com/HezaoHezao/poirot) | Difficulty: Intermediate | Time: 45 minutes | Basis: From the README Set the advanced feature flags so Poirot remembers past findings and improves its skills. **Who it is for:** A researcher who returns to the same subject over many sessions and wants earlier findings reused. **The problem:** Each new research session starts from zero, so earlier findings and working methods get lost. **The outcome:** A Poirot setup with memory recall, auto-consolidation, and skill evolution enabled in .env. ### Steps 1. Set up the project with `pip install -e ".[dev]"` after creating a virtual environment, as in the Quick Start. 2. Copy the sample config with `cp .env.example .env` and set DEEPSEEK_API_KEY. 3. Add the memory variables POIROT_MEMORY_USE=default, POIROT_MEMORY_PHASE2_ENABLED=true, and POIROT_MEMORY_PHASE2_TURNS=10 to .env. 4. Add the skill variables POIROT_SKILL_ENABLED=true, POIROT_SKILL_EVOLVE_ENABLED=true, and POIROT_SKILL_EVAL_ENABLED=true to .env. 5. Start Poirot with `poirot`, ask a research question on your topic, then type `/skill search` to browse skills. ### Prompt to paste into your agent ```text In the Poirot repo at , add the memory and skill variables from the README's Enable Advanced Features block to .env, then run `poirot` on the topic . Expected output: a session where memory recall and skill injection are active. Verify that the variables are set in .env before launching. ``` ### Caveat The README lists these flags as optional advanced features and sends full configuration details to USAGE.md, so confirm the behavior there 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/hezaohezao-poirot/enable-memory-and-skills-for-recurring-topic/ ## How to run research tasks inside a Docker sandbox on Windows with Poirot Repo: [HezaoHezao/poirot](https://github.com/HezaoHezao/poirot) | Difficulty: Advanced | Time: 60 minutes | Basis: From the README Route Poirot's tool execution through a Docker sandbox with WSL2 and enable specialist delegation. **Who it is for:** A developer on Windows with WSL2 who wants agent-run code and file writes isolated from the host. **The problem:** Letting an agent run commands and write files directly on a host machine exposes that machine to unintended changes. **The outcome:** A Poirot session where file writes and shell commands run in a Docker sandbox and outputs persist in the mount area. ### Steps 1. Set up the project with `pip install -e ".[dev]"` after creating a virtual environment, as in the Quick Start. 2. Confirm that Docker is running under WSL2 on the Windows machine. (inferred, check the README) 3. Add POIROT_SANDBOX_USE=Poirot.backend.agents.sandbox.docker.docker_sandbox_provider:DockerSandboxProvider and POIROT_SANDBOX_EXECUTOR=wsl to .env. 4. Add POIROT_MULTIAGENT_ENABLED=true to .env to allow specialist delegation. 5. Start Poirot with `poirot`, then save outputs under /mnt/poirot/user-data/ so they persist. ### Prompt to paste into your agent ```text In the Poirot repo at , set the Docker sandbox and WSL executor variables in .env, then run `poirot` with the task . Expected output: a completed task with its output files written under /mnt/poirot/user-data/. Verify that the output file exists in the mount path after the run. ``` ### Caveat The README says DockerPathGuard only allows writes under /mnt/poirot/user-data/, so files saved elsewhere inside the container may be lost when the container exits. Risk, credentials: Each specialist CLI (pi, codex, or claude) runs as its own process with its own LLM, so keep each provider's key in that tool's config and out of shared files. 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/hezaohezao-poirot/run-research-in-docker-sandbox/