Find your next open-source project

Search by name, purpose, or keyword — or paste a GitHub repository URL
All skills

vstorm-co/agenticos

Gives your self-hosted AI agents ready-made skills for coding, documents, business workflows and everyday office tasks.

  • 123 skills
  • 345 stars

The 5 most installed

  • Writes a follow-up message that addresses the customer's hesitation instead of repeating the same offer.

  • Checks a public document or web page against accessibility standards and lists what to fix.

  • Sets an agent's monthly spending budget and limits so mistakes stay cheap and busy months don't halt it.

  • Adds or changes what an agent can do, such as a new tool, knowledge source, web search or approval gate.

  • Turns a request like "I need an agent for X" into a design covering its job, permissions, knowledge and approvers.

The other 118 skills
  • Writes an agent's instructions so it does one job well, covering role, scope, refusals, sources and tone.

  • Changes the agent specification format, such as adding, renaming or validating fields, while keeping existing agents working.

  • Tests an agent before publishing it, suggesting questions to ask, failures to watch for and fixes to make.

  • Answers questions about AgenticOS features, setup and API from the product's own documentation.

  • Changes the PostgreSQL database schema with Alembic, including tables, columns, indexes, constraints and data backfills.

  • Writes the factual narrative behind an anti-money-laundering monitoring alert for an analyst to review.

  • Answers integration questions with a working request example and diagnoses problems from HTTP status codes.

  • Tells a patient what to bring, stop, fast for or arrange before a listed appointment.

  • Records a technical decision along with the options rejected and the cost accepted.

  • Builds polished, reusable apps, reports and decision tools with a stable published link.

  • Classifies a reported equipment fault, sets the response priority and dispatches it with the right details.

  • Writes or extends backend Python tests, including unit, database, API and bug-fix tests, and helps fix failing coverage.

  • Adds or debugs work that runs outside a request, such as document ingestion, syncs, scheduled jobs and reports.

  • Turns a bug report into something reproducible or states precisely what information is missing.

  • Builds a complete AgenticOS solution from a business case, combining agents, knowledge, skills, workflows and tables.

  • Assembles and checks a document bundle against a court's filing requirements.

  • Sets up or debugs chat bots on Telegram, Slack, Mattermost, Teams or Discord and routes messages through an agent.

  • Answers procedural questions from published information and routes everything else to the right place.

  • Turns a first notice of loss into a structured, complete insurance claim record.

  • Writes a client letter explaining where their matter stands, what it costs and what happens next.

  • Answers staff questions from SOPs and guidelines with citations, and refuses when the procedure is silent.

  • Applies your organization's review rules and flags what a code review must not let through.

  • Matches a documented episode to the right billing code and flags what the notes do not support.

  • Acknowledges, classifies and routes a complaint within the regulatory time limit.

  • Tracks a complaint through its statutory stages so no escalation right is missed.

  • Assembles the entity and relationship picture needed to decide a conflict of interest.

  • Retrieves the approved contract clause, its variants and the fallback position.

  • Summarises an application pack into only the facts a credit decision needs.

  • Explains an outage to customers honestly without leaking internals or blaming a person.

  • Computes and monitors the key dates of a matter and escalates before they pass.

  • Checks a finished deck for a clear storyline, sourced numbers, readability and a clear ask.

  • Plans what a presentation should say before slides are built, including audience, answer and storyline.

  • Detects a shipment going wrong and acts before the customer notices.

  • Judges whether a dependency upgrade is safe and says what to test.

  • Installs, verifies and updates a local or single-server AgenticOS deployment using the repository's scripts.

  • Runs a development workflow from design and plan through a merged PR, then writes tests and code.

  • Prepares an AgenticOS fork for platform changes, its own images and upstream updates.

  • Rewrites a discharge summary for the patient without changing any clinical fact.

  • Sorts documents for relevance and privilege, and never decides privilege on its own.

  • Produces polished papers, LaTeX documents, editable slides and exported reports with citations and build checks.

  • Extracts the operative terms of an agreement and flags what deviates from the standard.

  • Carries a large job through planning, specialists and background tasks without losing decisions or faking completion.

  • Writes or fixes Playwright end-to-end tests that cross the whole stack.

  • Logs, clarifies and routes a request for information within the statutory time limit.

  • Builds or changes Next.js pages, views and data-driven features, including stores, permissions and localized copy.

  • Helps you choose how to adopt AgenticOS, whether from a repo, fork, local install or existing deployment.

  • Checks a grant application against published criteria and lists what is still missing.

  • Verifies a delivery against the order and records problems before it is signed for.

  • Writes up an incident so the next person can act on it, not just read it.

  • Turns an account of a patient-safety incident into a complete, factual, blame-free report.

  • Turns an alert into a severity, an owner and a first hypothesis within minutes.

  • Turns stock levels into a few decisions with a lead time attached.

  • Checks an identity pack for completeness and consistency and lists what blocks onboarding.

  • Turns a maintenance backlog into a sequence that fits the production plan.

  • Captures a new legal matter completely and runs conflict and scope questions first.

  • Connects MCP servers, adds them to the catalog and debugs missing tools or failing OAuth flows.

  • Compares medication lists from two sources and surfaces every discrepancy for a pharmacist.

  • Produces meeting minutes that record decisions, actions and dissent rather than the discussion.

  • Picks a model for an agent by quality, speed, cost and data location, and explains why.

  • Drives a month-end close through its dependencies and reports what is blocking it.

  • Answers a new engineer's questions from the repository's own docs, citing the file path.

  • Operates an AgenticOS organization: builds and publishes agents, diagnoses runs and spend, and manages knowledge, routines and workflows.

  • Answers where-is-my-order questions from order and carrier records and says what happens next.

  • Decides which patient messages an agent may answer, which need a clinician, and what it must never do.

  • Gates routes and services by permission, wires roles and grants, and debugs who can see what.

  • Builds or repairs an AgenticOS agent: capabilities, prompt, skills, knowledge, testing and publishing.

  • Configures routines, watches, event triggers, task reminders and suggestions without duplicating them.

  • Decides what belongs in AgenticOS context versus instructions, skills, memory, knowledge, projects and tasks.

  • Turns a procedure or a successful chat into a reusable skill with triggers, references and test cases.

  • Builds tables with typed columns, views and safe writes, and links records to workflows and apps.

  • Designs and checks workflow graphs with dependencies, entry points, API integrations and retries.

  • Rewrites a policy so a resident can act on it without changing its meaning.

  • Explains what changed at renewal, why the price moved and what the options are.

  • Fetches all pull request review comments and presents them in a structured format.

  • Creates or updates a pull request with a clear, consistent description.

  • Builds a readable HTML slide deck with a clear storyline, sensible slide density and working navigation.

  • Checks an artwork file against press requirements and says exactly what to fix.

  • Answers a bidder's clarification question correctly and publishes it to every bidder.

  • Writes product listing copy from the spec sheet without inventing any attribute.

  • Answers questions about rates, fees and terms from the current document and cites the version.

  • Explains what the production schedule can absorb and what a change would cost.

  • Writes or fixes documentation pages, links, diagrams and navigation, and keeps docs builds passing.

  • Reviews a diff against the repository's own conventions in the order a maintainer would.

  • Reads an agent built with Pydantic AI, maps it to a draft and reports what did not translate.

  • Records a quality deviation, contains it and sets up the investigation without pre-judging it.

  • Ingests documents, runs semantic search, manages collections and connectors, and debugs retrieval problems.

  • Chooses the parser, chunking, retrieval and reranking setup that fits the documents at hand.

  • Handles a request to access records, covering identity checks, lawful basis, timescales and refusals.

  • Assembles a structured referral letter from the record and flags every field the notes cannot fill.

  • Explains when a refund is given without asking, when it needs approval and how to say no.

  • Compares two versions of a rulebook and reports what changed and who it affects.

  • Turns merged changes into release notes users can act on, grouped by what they mean.

  • Takes an agent through datasets, evals, published versions, staging, promotion and rollback.

  • Produces a source-grounded research brief or comparison with traceable claims, dates and reconciled contradictions.

  • Resolves a merge conflict in CHANGELOG.md by detecting the file's format and folding entries together.

  • Decides a return or exchange, flags cases needing a person, and refuses with a clear next step.

  • Publishes a private HTML walkthrough that maps a pull request's diff onto the system architecture.

  • Writes a reply to a public review that reassures the next reader instead of arguing with the reviewer.

  • Writes a runbook that someone woken at 3am can follow step by step.

  • Logs a near miss or unsafe condition so it produces a follow-up action, not just a statistic.

  • Writes an outage notice telling residents what is affected, what to do, and when service returns.

  • Prepares a shift handover so the next shift starts informed instead of discovering problems.

  • Answers questions about duties, delivery windows and restrictions for a destination accurately.

  • Rewrites prose into ASD-STE100 Simplified Technical English so docs read clear and human.

  • Detects a late purchase order, chases it with the right ask, and escalates on a schedule.

  • Raises a supplier nonconformance backed by evidence the supplier cannot dispute.

  • Answers from the correct revision of a drawing or manual and states which revision it used.

  • Writes a time-entry narrative a client will pay without questioning it.

  • Classifies a disputed transaction and routes it down the right path with the right evidence.

  • Stores, reads or rotates credentials and checks code for plaintext keys leaking into logs or responses.

  • Recognises high-value or at-risk customers and changes how their case is handled.

  • Designs and uses typed tables to store an automation's records, state, notes and reports.

  • Explains and compares ideas through diagrams, interactive models, charts, maps and original graphics.

  • Runs a staged multi-agent code review covering scope, correctness, security, quality and verification.

  • Takes a machine fault from symptom to the documented repair procedure, safely.

  • Builds and verifies automations and API integrations, including tools not in the MCP gallery.