Tool Design skill
Design agent-facing tool contracts with clear inputs, outputs, selection guidance and recoverable errors.
Community · Muratcan Koylan| Author | Muratcan Koylan |
|---|---|
| Licence | MIT |
| Source | muratcankoylan/Agent-Skills-for-Context-Engineering · SKILL.md on GitHub |
| Works in | Claude Code · Codex, Gemini CLI, Cursor, Copilot |
| Read from the source on |
The tool-design skill improves the interface between an agent and the tools it can call.
It treats each tool as a contract: when to select it, which inputs it accepts, what it returns and how a failed call can recover. It does not implement the tool backend.
From the skill's SKILL.md: ## Gotchas
What it does
- Audit overlapping names and ambiguous descriptions.
- Define parameters, response formats and error recovery.
- Test whether the agent selects and uses the intended tool.
What you get
- Tool contracts, revised descriptions and a validation plan.
When to use it
- An agent chooses the wrong tool or repeatedly supplies invalid arguments.
When not to use it
- Choosing an entire system architecture or merging unrelated operations into one oversized tool.
What it needs
- A file-capable agent and the actual tool schemas, behavior and failure cases.
- The optional Python generator uses templates and heuristic scoring. It does not prove that descriptions improve real tool selection.
Files in the skill
| File or folder | What it is for |
|---|---|
SKILL.md | Tool-contract guidance, examples and design pitfalls. |
references/best_practices.md | A Python description generator and reference guides for design and consolidation. |
scripts/description_generator.py | A Python description generator and reference guides for design and consolidation. |
Licence as found in the repository on 2026-11-20: MIT.
Example prompts
Type one of these after installing; naming the skill makes sure it is used.
- Rewrite this tool contract with explicit defaults and recoverable errors.
- Audit these overlapping tool names before proposing consolidation.
- Design concise and detailed responses for the same search operation.
How to install tool-design
Preserve references and examples
Copy the complete folder. The context-engineering marketplace installs this skill together with the collection; use a folder install for one skill.
No external service is required for the text guidance or Python templates. Testing real selection needs your agent and tool implementation; hosted upload and API execution are not verified here.
A skill is a folder with a SKILL.md file. You install it once; the agent reads its short description at the start of each session and loads the full instructions only when a request matches. Pick the place you use:
Claude Code: from the plugin marketplace
- Start Claude Code in a terminal with
claude. - Add the marketplace (once per computer):
/plugin marketplace add muratcankoylan/Agent-Skills-for-Context-Engineering - Install the plugin. Claude Code opens its details first: check what it will add (skills, hooks, MCP servers), then choose whether it is for you, for everyone on this repository, or for you in this repository only.
/plugin install context-engineering@context-engineering-marketplace - Type / and look for
/context-engineering:tool-design, or simply ask for the task.
From your shell instead of a session:
claude plugin marketplace add muratcankoylan/Agent-Skills-for-Context-Engineering
claude plugin install context-engineering@context-engineering-marketplaceThe plugin installs as a whole, so other skills come with it, including context-fundamentals, context-degradation, context-compression, context-optimization, self-managed-context, multi-agent-patterns, long-horizon-prompting, memory-systems, filesystem-context, hosted-agents, evaluation, advanced-evaluation, harness-engineering, project-development, bdi-mental-states, latent-briefing, self-improvement-loops.
Claude Code: copy the folder into your skills folder
Put the skill's folder in ~/.claude/skills/ to use it in every project on this computer, or in .claude/skills/ inside a repository to share it with everyone who works there (commit it).
git clone --depth 1 https://github.com/muratcankoylan/Agent-Skills-for-Context-Engineering.git
mkdir -p ~/.claude/skills
cp -R Agent-Skills-for-Context-Engineering/skills/tool-design ~/.claude/skills/Claude Code notices new or changed skills in these folders during the session. If the skills folder did not exist when the session started, run /reload-skills once. Then type /tool-design or just describe the task.
On Windows, run the commands in Git Bash or WSL, or copy the folder into the same place with File Explorer.
Codex, Gemini CLI, Cursor and GitHub Copilot
These agents read the same SKILL.md folders. The one folder all four accept inside a project is .agents/skills/, so this copies the skill there:
git clone --depth 1 https://github.com/muratcankoylan/Agent-Skills-for-Context-Engineering.git
mkdir -p .agents/skills
cp -R Agent-Skills-for-Context-Engineering/skills/tool-design .agents/skills/| Agent | Project folder | Your user folder | Notes |
|---|---|---|---|
| OpenAI Codex | .agents/skills/ | ~/.agents/skills/ | Type $ or run /skills to pick a skill; Codex also chooses one when your task matches. |
| Gemini CLI | .agents/skills/.gemini/skills/ | ~/.agents/skills/~/.gemini/skills/ | Or install from Git with gemini skills install <git URL> --path <folder> --consent. Google replaced Gemini CLI with Antigravity CLI for free-tier and Google One users on 18 June 2026; Antigravity uses .agents/skills/ in a project and ~/.gemini/config/skills/ for you. |
| Cursor | .agents/skills/.cursor/skills/ | ~/.agents/skills/~/.cursor/skills/ | Type / in Agent chat to pick a skill. Cursor also reads Claude Code's .claude/skills/ folders. |
| GitHub Copilot | .github/skills/.agents/skills/.claude/skills/ | ~/.copilot/skills/~/.agents/skills/ | Works in Copilot CLI, the cloud agent, and agent mode in VS Code, Visual Studio and JetBrains (IDE support depends on the version). Or run gh skill install OWNER/REPOSITORY SKILL (public preview, GitHub CLI 2.90.0 or later). |
Sources for these steps
Checked against each vendor's documentation on 2026-11-14:
A contract that supports the next action
Names and descriptions should help the agent choose among real alternatives. State defaults, input constraints and output structure rather than relying on a suggestive name.
Offer concise or detailed output when the task needs it. Error messages should make a corrective action possible instead of inviting repeated blind retries.
Validate consolidation against the workload
Reducing tool count is a design option, not a performance guarantee. Compare success, argument errors and recovery on the same representative tasks.
The bundled generator can draft descriptions. Its heuristic checks do not verify backend behavior, permissions or whether an agent understands the contract.
Questions about tool-design
Does it build an MCP server?
No. It can guide a tool contract, but transport, authentication and backend execution still need implementation.
Should every tool be consolidated?
No. Consolidate overlapping jobs where useful; keep unrelated operations separate and test the resulting interface.
What should an error response contain?
Explain what failed and which input or state must change before retrying, without exposing sensitive implementation details.
Are the example quality scores measured agent performance?
No. They check text patterns and metadata. Validate tool selection and recovery with representative agent runs.
More AI skills
Related skills
Dev-Browser skill
Community · Sawyer HoodAutomate Chrome with persistent named pages, Puppeteer scripts and a local command-line service.Agent Evaluation skill
Community · Muratcan KoylanPlan agent regression suites with deterministic checks, separate quality dimensions and versioned baselines.Context Degradation skill
Community · Muratcan KoylanDiagnose long-context failures and plan recovery using verified facts and workload-specific comparisons.Advanced Evaluation skill
Community · Muratcan KoylanDesign LLM judges with evidence-based rubrics, swapped pairwise comparisons and human calibration.Brainstorming skill
Community · Jesse Vincent / Prime RadiantTurn an undeveloped idea into an agreed design before writing implementation code.Writing Plans skill
Community · Jesse Vincent / Prime RadiantConvert an approved design into reviewable engineering tasks with interfaces and verification.