- Home
- Skills
- Dragon1086
- Claude Skills
- Tool Advisor
tool-advisor_skill
- Shell
3
GitHub Stars
1
Bundled Files
2 months ago
Catalog Refreshed
4 months ago
First Indexed
Readme & install
Copy the install command, review bundled files from the catalogue, and read any extended description pulled from the listing source.
Installation
Preview and clipboard use veilstrat where the catalogue uses aiagentskills.
npx veilstrat add skill dragon1086/claude-skills --skill tool-advisor- SKILL.md17.0 KB
Overview
This skill discovers your full tool environment and amplifies prompts with capability awareness to help you choose the best tool composition. It never executes actions; instead it enriches the model's context and suggests non‑binding tool chains. The output is a disciplined, six‑phase advisor flow that ends with a concise Quick Action table.
How this skill works
It scans known locations and reports native tools, installed skills, plugins, agents, MCP servers, and common CLI dev tools. It classifies the task, maps relevant discovered capabilities to that task, and then proposes 1–3 concrete tool chains as options with tradeoffs. Each run produces a standardized template covering Environment, Task Profile, Relevant Capabilities, Suggested Approaches, Gaps, Performance Tips, and a Quick Action table.
When to use it
- Before starting code changes to map available tooling and avoid surprises
- When you want the model to be capability-aware for workflow planning
- To get non-binding, concrete tool chains that match installed skills/plugins
- When coordinating multi-step or multi-agent work and you need explicit tradeoffs
- For onboarding or auditing a developer environment to list gaps
Best practices
- Treat suggestions as optional — the model's judgment still rules
- Do not expect the skill to execute commands; it only recommends first steps
- Provide a clear "Done when" line or answer up to 3 clarifying questions if asked
- Use the Quick Action table to copy/paste starter commands into your shell or agent
- Run the full six phases for medium/large tasks; collapse for very small tasks
Example use cases
- Select an optimal tool chain to run tests, lint, and CI checks before a release
- Discover installed MCP servers and suggest which to query for framework docs
- Compose a safe, methodical approach to refactor an auth module without executing changes
- Audit a new developer VM to list missing dev tools and recommend installs
- Generate a Quick Action table of first commands to hand to an executor agent
FAQ
No. It never executes actions. It only scans known config locations and recommends commands or tool chains for you or an executor to run.
What happens if a useful tool is missing?
It lists missing-but-useful tools in a "Could be useful" section and explicitly states the task is doable without them; installation is optional and requires your approval.