Package repeatable instructions and supporting resources so agents know how to perform specialised work—while VDF AI keeps tool permissions, model access, approvals, and runtime controls authoritative.
name: web-research
description: Research current questions with sources
allowed-tools: web_search web_crawler
# Web research
Load the complete instructions only when relevant.
Use assigned tools; never expand permissions.Each included Skill gives agents a focused method, clear outputs, and honest boundaries. Search the catalog by job to be done; permissions are still configured independently.
8 included skills
Research current questions against live public sources and return a concise, cited answer.
v1.0 KnowledgeSearch indexed enterprise systems and answer only from the connected knowledge available to the user.
v1.0 DocumentsRead attached or linked business files and answer from the contents, including scans that require OCR.
v1.0 DocumentsTurn complete, reviewed content into downloadable documents, PDFs, presentations, spreadsheets, or Markdown.
v1.0 AnalysisCompare alternatives against explicit criteria, expose trade-offs, and finish with one recommendation.
v1.0 AnalysisProfile a dataset, identify quality problems and anomalies, and deliver a concise review with a chart.
v1.0 Governed actionGuide external changes through a preview-and-agree contract before an authorised write tool is used.
v1.0 Governed actionTurn notes, bugs, meetings, or documents into implementation-ready Jira issues, GitHub issues, or PR descriptions.
No Agent Skills match this search. Try a broader term or another category.
Agent Skill gives the agent a repeatable method. Tool permissions stay separate, and the capability can run inside a governed multi-agent Network.
Build Skills as operational assets: validate their structure, align tools with least privilege, test activation, and preserve reviewable versions.
Start in the editor, upload a Skill ZIP, or import supported skill content from an Agent Plugin.
Check the bundle structure and metadata, then resolve every reported issue before publishing.
Assign the Skill to an agent and separately grant only the tools its procedure can use.
Try explicit invocation and representative requests, then inspect activation and file-read events.
Preserve a reviewable snapshot, monitor usage, and update the Skill as the workflow changes.
The public catalog describes VDF’s included system Skills. Inside Agent Hub, users can also create private procedures for the work unique to their organisation.
Use VDF-maintained procedures for common research, knowledge, document, analysis, and action workflows.
Create in the editor or upload a compatible ZIP, validate the bundle, and assign it to agents without publishing proprietary instructions.
Agent Hub owns and manages Skills. VDF AI Networks resolves them under the caller’s authorisation and attaches them to model-driven nodes without turning a Skill into a new node type.
Add Skills during the Tools & Skills step. VDF reports missing tool bindings so builders can close the capability gap deliberately.
LLM Agent, Aggregator, and Router nodes can inherit Skills from a source agent or receive a node-specific selection.
Separating these layers makes an agent easier to govern, reuse, test, and explain.
| Layer | What it defines | What it does not do |
|---|---|---|
| Agent | Identity, model, role, and runtime configuration | Does not replace a reusable work method |
| Skill | Procedure, decision guidance, and supporting resources | Does not grant tools or connector access |
| Prompt | The immediate request and conversation context | Does not provide a maintained operational asset |
| Tool | A callable capability such as search or generation | Does not decide how the wider job should be done |
| MCP | A protocol for exposing tools and context | Does not define the specialist procedure |
| Network | Routing and orchestration across model-driven nodes | Does not collapse governance into one prompt |
VDF follows the open Agent Skills directory model while keeping execution decisions inside the platform boundary your administrators control.
Agent Skills are portable bundles of instructions and optional supporting resources that teach an AI agent how to perform a specialized workflow. On VDF AI, Skills remain separate from tool permissions, model access, and runtime policy.
A Skill describes how work should be done. A tool provides a capability such as searching, parsing, generating, analysing, or writing. Assigning a Skill never grants its tools; those permissions are configured separately on the agent.
Yes. Users can create or upload private Skills, validate their structure, bind them to agents, and maintain versions. Private user Skills are visible only within their authorised product context and are not published to this public catalog.
Yes. Skills can be assigned to model-driven nodes such as LLM Agent, Aggregator, and Router nodes. A node can inherit Skills from its source agent or define its own selection, and unresolved Skills are reported in the run output.
Not in the current VDF AI runtime. A Skill bundle can store scripts alongside instructions, references, and assets, but this environment exposes files for controlled reading rather than arbitrary script execution.
Start with eight included Skills, create private procedures for your organisation, and keep every tool and external action inside governed platform controls.