# AI use cases for Awesome Deep Seek Harness Plugins (Zhiyuan-Fan/Awesome-DeepSeek-Harness-Plugins) Repo: [Zhiyuan-Fan/Awesome-DeepSeek-Harness-Plugins](https://github.com/Zhiyuan-Fan/Awesome-DeepSeek-Harness-Plugins) | Use cases: 4 | Page: https://dedupedai.com/use-cases/zhiyuan-fan-awesome-deepseek-harness-plugins/ A curated, daily-updated directory of public plugins, tools, clients, runtimes, and references for the DeepSeek Harness agent harness. This file holds every use case of this repo. Each one starts with a "## " heading and ends with its URL. - Run DSH on a local Ollama model with cloud fallback: https://dedupedai.com/use-cases/zhiyuan-fan-awesome-deepseek-harness-plugins/local-ollama-model-with-cloud-fallback/ - Review DSH turns with a second model's advisory notes: https://dedupedai.com/use-cases/zhiyuan-fan-awesome-deepseek-harness-plugins/second-model-advisory-notes-on-dsh-turns/ - Assign different models to DSH planner and subagent roles: https://dedupedai.com/use-cases/zhiyuan-fan-awesome-deepseek-harness-plugins/separate-models-for-planner-and-subagent-roles/ - Fill missing metadata for custom DSH model endpoints: https://dedupedai.com/use-cases/zhiyuan-fan-awesome-deepseek-harness-plugins/fill-missing-metadata-for-custom-model-endpoints/ ## How to run DSH on a local Ollama model with cloud fallback with Awesome Deep Seek Harness Plugins Repo: [Zhiyuan-Fan/Awesome-DeepSeek-Harness-Plugins](https://github.com/Zhiyuan-Fan/Awesome-DeepSeek-Harness-Plugins) | Difficulty: Intermediate | Time: 1 hour | Basis: From the README Add an Ollama provider to DSH so local models handle work, with cloud models as a backup. **Who it is for:** A developer who wants DSH to use models on their own machine and has an Ollama install ready. **The problem:** DSH needs a model provider, and a local model can keep prompts on your machine, but a local model may be unavailable at times. **The outcome:** DSH sends requests to Ollama models, checks their health, and falls back to a cloud model when needed. ### Steps 1. Open the dsh-local-ai repository (https://github.com/PerryLink/dsh-local-ai) and read its README for install steps. 2. Check that Ollama is running and that the model names you plan to use are available to it. 3. Install the plugin into DSH following the repository's instructions. 4. In DSH, select the Ollama provider and confirm the health check reports the local server as available. 5. Configure the cloud fallback provider, then send a test request while the local server is stopped to confirm the fallback is used. ### Prompt to paste into your agent ```text Read the README of the repository at (dsh-local-ai) and list its install steps, the Ollama settings it needs, and how its cloud fallback is configured. Expected output: a numbered setup checklist for a DSH install on . Verify that every step is stated in the README and mark anything that is not as unconfirmed. ``` ### Caveat The README lists the Ollama provider and fallback features but gives no install commands, so confirm the exact setup steps in the plugin repository before relying on them. Risk, credentials: Cloud fallback requires provider credentials, so store API keys in the location the plugin documents and do not paste them into shared prompts or logs. 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/zhiyuan-fan-awesome-deepseek-harness-plugins/local-ollama-model-with-cloud-fallback/ ## How to review DSH turns with a second model's advisory notes with Awesome Deep Seek Harness Plugins Repo: [Zhiyuan-Fan/Awesome-DeepSeek-Harness-Plugins](https://github.com/Zhiyuan-Fan/Awesome-DeepSeek-Harness-Plugins) | Difficulty: Intermediate | Time: 1 hour | Basis: From the README Have a separate model flag issues in each DSH turn without blocking the main agent. **Who it is for:** A developer running DSH on code or content tasks who wants a second opinion on each step without manual review of every turn. **The problem:** A single agent can make mistakes that go unnoticed, but a manual check on every turn is slow. **The outcome:** A configured reviewer model adds severity-ranked notes to each turn, and the primary agent keeps its actions. ### Steps 1. Open the dsh-advisor repository (https://github.com/omdsh-dev/dsh-advisor) and read its README to confirm how the reviewer model is set up. 2. Pick a second model endpoint you can access and note its provider details. 3. Install the plugin into DSH following the repository's instructions. 4. Configure the separate reviewer model in the plugin settings. 5. Run a short task and check that the advisory notes appear ranked by severity and that the primary agent's actions were not changed. ### Prompt to paste into your agent ```text Read the README of the repository at (dsh-advisor) and summarize how the reviewer model is configured and how advisory notes are presented. Expected output: a short setup summary for a DSH project at . Verify that the README states the notes do not approve or reject the primary agent's actions before you report that behavior. ``` ### Caveat The README says advisory notes do not approve or reject actions, so this helps review but does not enforce a gate on the primary agent. Risk, credentials: The reviewer model needs its own endpoint credentials, so keep them out of shared configuration 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/zhiyuan-fan-awesome-deepseek-harness-plugins/second-model-advisory-notes-on-dsh-turns/ ## How to assign different models to DSH planner and subagent roles with Awesome Deep Seek Harness Plugins Repo: [Zhiyuan-Fan/Awesome-DeepSeek-Harness-Plugins](https://github.com/Zhiyuan-Fan/Awesome-DeepSeek-Harness-Plugins) | Difficulty: Intermediate | Time: 1 hour | Basis: From the README Send planner and subagent roles to different configured models inside DSH. **Who it is for:** A developer whose DSH tasks use a planner plus several subagents and who wants each role on a model that fits it. **The problem:** Every role in a multi-agent run uses the same model by default, which may not suit planning and execution equally well. **The outcome:** The planner and each subagent run on the model you assign, and the active routing summary is visible in DSH Web. ### Steps 1. Open the dsh-role-router repository (https://github.com/SnowAmberX/dsh-role-router) and read its README for install and role configuration. 2. List the models you have configured in DSH and choose one for the planner and one for subagents. 3. Install the plugin into DSH following the repository's instructions. 4. Map the planner role and subagent roles to their chosen models in the plugin settings. 5. Open DSH Web and confirm the routing summary shows the mapping you set. ### Prompt to paste into your agent ```text Read the README of the repository at (dsh-role-router) and list the role names it supports and how a role is mapped to a model. Expected output: a mapping table for with planner and subagent roles. Verify that the README confirms the routing summary is shown in DSH Web before you include that step. ``` ### Caveat The README names planner and subagent roles, but the repo also lists dsh-plugin-subagent-director for per-subagent provider and model choices, so compare both before picking one. 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/zhiyuan-fan-awesome-deepseek-harness-plugins/separate-models-for-planner-and-subagent-roles/ ## How to fill missing metadata for custom DSH model endpoints with Awesome Deep Seek Harness Plugins Repo: [Zhiyuan-Fan/Awesome-DeepSeek-Harness-Plugins](https://github.com/Zhiyuan-Fan/Awesome-DeepSeek-Harness-Plugins) | Difficulty: Intermediate | Time: 45 minutes | Basis: From the README Auto-complete context, output, and modality fields for custom models from a models.dev index. **Who it is for:** A developer who added custom model endpoints to DSH and found that context limits or modalities are blank or wrong. **The problem:** Custom models often lack the metadata DSH needs, and entering it by hand for each model is tedious. **The outcome:** Missing context, output, display-name, and modality fields are filled from a bundled models.dev index that you can refresh. ### Steps 1. Open the dsh-model-meta-autofill repository (https://github.com/QJAG1024/dsh-model-meta-autofill) and read its README for install steps. 2. List the custom model endpoints in DSH that have blank metadata fields. 3. Install the plugin into DSH following the repository's instructions. 4. Run the autofill and review each filled field against the provider's own documentation. 5. Refresh the bundled models.dev index when you add new models. ### Prompt to paste into your agent ```text Read the README of the repository at (dsh-model-meta-autofill) and list which metadata fields it fills and how it refreshes its index. Expected output: a field list and refresh steps for a DSH setup at . Verify that the README names the bundled models.dev index before you say the index is refreshable. ``` ### Caveat The filled values come from a models.dev index, so check them against each provider's own documentation before relying on context or output limits. 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/zhiyuan-fan-awesome-deepseek-harness-plugins/fill-missing-metadata-for-custom-model-endpoints/