create-custom-agent
by dotnetcreate-custom-agent helps you create VS Code custom agent files (.agent.md) for specialized AI personas with tools, instructions, and handoffs. Use it to scaffold new agents, set tool limits, and define agent-to-agent workflows for Skill Authoring.
This skill scores 74/100, which means it is acceptable to list and should be useful for users who want to create VS Code custom agents without starting from a blank prompt. The repository gives enough workflow and usage guidance to support installation decisions, though users should note some missing supporting assets and a few placeholder markers that limit polish and depth.
- Clear triggerability for creating new VS Code custom agent files, with explicit when-to-use guidance and a matching frontmatter description.
- Good operational structure: inputs, workflow steps, and handoff/tool configuration are documented, making the skill easier for agents to execute than a generic prompt.
- Strong install-decision value for agent authors because it distinguishes custom agents from instruction/prompt files and covers workspace-shared or user-profile agents.
- No scripts, references, or companion files are included, so users must rely on the prose alone for implementation details.
- Placeholder markers and lack of an install command suggest the skill is more guidance-focused than turnkey automation.
Overview of create-custom-agent skill
The create-custom-agent skill helps you create VS Code custom agent files (.agent.md) for specialized AI roles. It is best for people who need a reusable agent setup with explicit tools, instructions, and handoffs rather than a one-off prompt. If you are building planner, reviewer, or task-specific agents, this skill gives you a structured starting point for the create-custom-agent for Skill Authoring workflow.
What this skill is for
Use create-custom-agent when you want to define how an agent should think, what tools it may use, and what happens next after it finishes. It is aimed at agent scaffolding, not general prompt writing.
Who should install it
Install the create-custom-agent skill if you work in VS Code, want .agent.md files in your repo or user profile, or need repeatable agent behavior across tasks and teammates.
What makes it useful
The main value is consistency: the skill turns vague role ideas into a concrete agent file with frontmatter, tool limits, and handoff logic. That reduces guesswork when setting up custom agent workflows.
How to Use create-custom-agent skill
Install the create-custom-agent skill
Run the install step from your skills manager, for example: npx skills add dotnet/skills --skill create-custom-agent. After install, open the skill files in .agents/skills/create-custom-agent and use SKILL.md as the primary guide.
Turn a rough idea into a usable prompt
For better create-custom-agent usage, provide these inputs up front: agent name, short description, intended role, allowed tools, and any handoff target. A weak request like “make me an agent” usually produces a generic result; a stronger one is “create a code-reviewer agent for C# PRs, limited to read, search, and comment tools, with a handoff to a fixer agent.”
Read these files first
Start with SKILL.md, then inspect any linked repo context the skill references in the file itself. In this repository there are no helper folders, so the fastest path is to read the workflow, inputs, and constraint sections closely before you generate your own .agent.md.
Use a practical workflow
Draft the persona first, then decide tool scope, then define handoffs last. That order matters because tool access and follow-up routing affect the agent’s actual behavior more than naming or branding.
create-custom-agent skill FAQ
Is create-custom-agent only for VS Code agents?
Yes, this skill is centered on VS Code custom agents and .agent.md authoring. If you need .instructions.md or .prompt.md, this is the wrong install.
How is this different from a normal prompt?
A prompt is ephemeral; a custom agent is reusable and constrained. The create-custom-agent guide is useful when you need a durable agent definition with tool permissions and workflow handoffs, not just a single chat instruction.
Is it beginner-friendly?
Yes, if you already know the job you want the agent to do. It is less beginner-friendly when the role, tools, or handoff path are still unclear, because those choices drive the quality of the generated file.
When should I not use it?
Do not use create-custom-agent if you are editing an existing agent directly, making an instruction file, or only need a disposable prompt for one task.
How to Improve create-custom-agent skill
Give the skill stronger source inputs
The biggest quality boost comes from a specific agent brief: task domain, success criteria, allowed tools, forbidden tools, and the next agent in the chain. The clearer those inputs are, the less cleanup you need after generation.
Watch for common failure modes
The usual mistakes are vague personas, overly broad tool sets, and handoffs that do not match the real workflow. If the output feels generic, your prompt likely described a role instead of a job.
Iterate after the first draft
Use the first .agent.md as a draft, then tighten description text, tool limits, and handoff conditions based on real use. For create-custom-agent skill output, small constraint edits usually improve behavior more than rewriting the whole file.
Validate against your repository needs
Check whether the agent fits your repo’s conventions, security boundaries, and team workflow before publishing it. The best create-custom-agent results are the ones that match an actual operating context, not just a neat persona.