# AI use cases for Skill Corpus (EverMind-AI/SkillCorpus) Repo: [EverMind-AI/SkillCorpus](https://github.com/EverMind-AI/SkillCorpus) | Use cases: 4 | Page: https://dedupedai.com/use-cases/evermind-ai-skillcorpus/ SkillCorpus is an open-source Python pipeline that collects SKILL.md files from public repositories, curates and scores them, and provides retrieval models and evaluation tools for matching skills to agent tasks. This file holds every use case of this repo. Each one starts with a "## " heading and ends with its URL. - Fetch task-specific skills from SkillHub for an agent: https://dedupedai.com/use-cases/evermind-ai-skillcorpus/fetch-task-skills-from-skillhub-api/ - Install the Skill Corpus plugin in your agent host: https://dedupedai.com/use-cases/evermind-ai-skillcorpus/install-skillcorpus-plugin-in-agent-host/ - Build a private skill corpus from your own registry: https://dedupedai.com/use-cases/evermind-ai-skillcorpus/build-custom-skill-corpus-from-own-sources/ - Self-host the Skill Corpus retrieval models: https://dedupedai.com/use-cases/evermind-ai-skillcorpus/self-host-skill-retrieval-models/ ## How to fetch task-specific skills from SkillHub for an agent with Skill Corpus Repo: [EverMind-AI/SkillCorpus](https://github.com/EverMind-AI/SkillCorpus) | Difficulty: Intermediate | Time: 15 minutes | Basis: From the README Search the hosted SkillHub catalog with a plain task description and inject the matching SKILL.md text into a prompt. **Who it is for:** A developer building an agent who needs step-by-step procedures for a specific task, such as extracting tables from PDFs, and wants them without writing the prompt by hand. **The problem:** Generic model knowledge often misses exact steps, edge cases, or supporting scripts for a narrow task. **The outcome:** The agent receives the full skill_md instructions for the best-matching skills before it starts the task. ### Steps 1. Run `curl "https://skillhub.evermind.ai/openapi/v1/skills?q=extract+tables+from+a+PDF"` and note the `id` of a relevant result. 2. Fetch the `skill_md` for that `id` using the read tier described in docs/integrations.md. 3. Insert the returned skill_md text into your agent's system or task prompt. 4. Run your agent on the task and check that it follows the injected steps. ### Prompt to paste into your agent ```text Using the SkillCorpus API at https://skillhub.evermind.ai/openapi/v1/skills, search for skills matching the task , pick the most relevant result, fetch its skill_md, and output the injected prompt text. Verify that the returned skill_md is non-empty and that the skill license is shown in the metadata before using it. ``` ### Caveat The README says each skill keeps its upstream license, so check the license field of any skill before reusing its content in your product. Dual-use notice: only use this on software, sites and systems you own or are authorized to test. 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/evermind-ai-skillcorpus/fetch-task-skills-from-skillhub-api/ ## How to install the Skill Corpus plugin in your agent host with Skill Corpus Repo: [EverMind-AI/SkillCorpus](https://github.com/EverMind-AI/SkillCorpus) | Difficulty: Beginner | Time: 15 minutes | Basis: From the README Let your agent install the Skill Corpus plugin for OpenClaw, Hermes, Raven, WorkBuddy, or DeepSeek Harness. **Who it is for:** A user of one of the five supported agent platforms who wants the agent to search Skill Corpus skills during normal work. **The problem:** Manually wiring a skill retrieval layer into each agent host takes setup time and per-host configuration. **The outcome:** The agent has a skill_search tool or automatic skill retrieval, using one shared skills directory across hosts. ### Steps 1. Open https://github.com/EverMind-AI/SkillCorpus/blob/main/skillcorpus_plugin/INSTALL.agent.md and confirm it matches your host. 2. Paste this line to your agent: Install Skill Corpus Plugins following https://github.com/EverMind-AI/SkillCorpus/blob/main/skillcorpus_plugin/INSTALL.agent.md 3. Choose a mode: on demand (the default, the agent calls skill_search) or mode: auto (search every turn). Do not enable both. 4. Review the five settings described in skillcorpus_plugin/ and check what leaves your machine on each turn. ### Prompt to paste into your agent ```text Read https://github.com/EverMind-AI/SkillCorpus/blob/main/skillcorpus_plugin/INSTALL.agent.md and install the SkillCorpus plugin for . Report which mode you configured (on demand or auto) and the path of the shared skills directory. Verify that a test query returns at least one skill. ``` ### Caveat The README says mode: auto for the Raven plugin is inert until Raven merges its upstream context_segments slot, so only the on-demand mode works there today. Risk, remote script: The install step has your agent follow instructions from a remote markdown file, so read the INSTALL.agent.md file before letting the agent run it. Dual-use notice: only use this on software, sites and systems you own or are authorized to test. 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/evermind-ai-skillcorpus/install-skillcorpus-plugin-in-agent-host/ ## How to build a private skill corpus from your own registry with Skill Corpus Repo: [EverMind-AI/SkillCorpus](https://github.com/EverMind-AI/SkillCorpus) | Difficulty: Advanced | Time: 2 hours | Basis: From the README Run the Skill Corpus pipeline on your own source registry, applying curation, safety, and license gates. **Who it is for:** A team that maintains its own internal SKILL.md files and wants to filter, score, and deduplicate them before giving them to agents. **The problem:** Raw skill collections mix duplicates, unsafe content, and unclear licenses, which makes them unreliable as agent context. **The outcome:** An exported corpus in parquet form with attachments and a dataset card, ready to serve to your agents. ### Steps 1. Clone the repository with `git clone https://github.com/EverMind-AI/SkillCorpus` (inferred from the repo URL). (inferred, check the README) 2. Point the aggregate step at your own source registry as described in the aggregate/ and curate/ modules. 3. Run the build chain with the cli in skillcorpus/cli.py, which the README says runs ingest, quality_pass, dedup_pass, license_audit, and export.corpus. (inferred, check the README) 4. Check the output with the stats command and the schema in docs/corpus-schema.md. ### Prompt to paste into your agent ```text Clone https://github.com/EverMind-AI/SkillCorpus, configure the aggregate step with the source registry at , run the build command in skillcorpus/cli.py, and report the exported corpus path and skill count. Verify that the export contains skills.parquet and the dataset card. ``` ### Caveat The README says LLM classification and quality scoring fall back to rules when no model endpoint is reachable, so results without a model endpoint may be less refined. Dual-use notice: only use this on software, sites and systems you own or are authorized to test. 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/evermind-ai-skillcorpus/build-custom-skill-corpus-from-own-sources/ ## How to self-host the Skill Corpus retrieval models with Skill Corpus Repo: [EverMind-AI/SkillCorpus](https://github.com/EverMind-AI/SkillCorpus) | Difficulty: Advanced | Time: 2 hours | Basis: From the README Serve the released bi-encoder and reranker behind your own /embed and /score endpoint. **Who it is for:** A platform engineer who must keep skill retrieval inside their own infrastructure instead of calling the hosted SkillHub API. **The problem:** Depending on a hosted endpoint may not fit data-control or deployment rules. **The outcome:** A local endpoint that encodes queries and skills and reranks top results for your agents. ### Steps 1. Load the demo corpus with `pip install pandas` and then read skills.parquet with pandas, or use the datasets library. 2. Run `pip install -r skillcorpus/match/requirements.txt` to install the serving dependencies. 3. Set EMBEDDING_MODEL and RERANKER_MODEL to the released checkpoints, then run `bash skillcorpus/match/scripts/run_server.sh`. 4. Call /embed and /score on the local endpoint and run your own encode, top-k, and rerank step. ### Prompt to paste into your agent ```text In the SkillCorpus repository, install skillcorpus/match/requirements.txt, start the server with run_server.sh using the released EverMind-AI bi-encoder and reranker checkpoints, then send a test query to /embed and /score. Report the top three skill ids for . Verify that the scores are returned for each candidate. ``` ### Caveat The README says this local endpoint is not a drop-in for the hosted /openapi/v1/skills API, so you must write your own selection logic over /embed and /score. Dual-use notice: only use this on software, sites and systems you own or are authorized to test. 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/evermind-ai-skillcorpus/self-host-skill-retrieval-models/