tool-budget-audit — Claude Code skill: context audit
tool-budget-audit is a Claude Code skill that measures context used by tool schemas and call results. It flags catalog bloat and mistaken tool choices.
MIT ★ 741 updated 2026-11-11
/plugin marketplace add MiniMax-AI/mcode-stack
/plugin install ronny@mcode-stackWhat tool-budget-audit does
tool-budget-audit measures how much context an agent spends on tool definitions, returned data, and extra model requests. It asks for tokenizer-based estimates and observations from real tasks, then identifies the main bottleneck.
The workflow starts by merging overlapping tools, clarifying names and return formats, and grouping tools by domain. Low-frequency tools can move to tool search, while filtering, aggregation, and parallel batches can move into programmatic calls. The output is a short report with measured figures, concrete tool names, and the next measurement to run. The companion checklist is checklists/tool-budget.md.
The audit also covers catalog fetches over the network, descriptions that reduce search recall, and name-based reminders or counters that stop firing after a tool moves into a program. End-to-end comparison uses harness-experiment.
What it helps with
- measure tokens spent on tool definitions
- compare returned data with model request count
- find tools that look similar to the agent
- decide which tools to merge or group
- choose between tool search and programmatic calls
Example prompts for the agent
- Audit every tool and MCP server in this project for context usage.
- Measure definitions, call counts, request counts, and result sizes on two representative tasks.
- Write the short report with the bottleneck, named tools, and the next measurement.
How to install tool-budget-audit
1Plugin marketplace
/plugin marketplace add MiniMax-AI/mcode-stack/plugin install ronny@mcode-stackThis installs the whole ronny plugin with all its skills
2Skills CLI (Vercel)
npx skills add MiniMax-AI/mcode-stack --skill tool-budget-audit -a claude-code3Manually: put the files in place
~/.claude/skills/tool-budget-audit/SKILL.mdCopy the whole skill folder; for one project use .claude/skills/tool-budget-audit/
1Skills CLI (Vercel)
npx skills add MiniMax-AI/mcode-stack --skill tool-budget-audit -a codex2Manually: put the files in place
.agents/skills/tool-budget-audit/SKILL.mdFor all projects use ~/.agents/skills/
1Skills CLI (Vercel)
npx skills add MiniMax-AI/mcode-stack --skill tool-budget-audit -a cursor2Manually: put the files in place
.cursor/skills/tool-budget-audit/SKILL.mdCursor also reads .agents/skills/ and .claude/skills/
1GitHub CLI
gh skill install MiniMax-AI/mcode-stack stacks/ronny/skills/tool-budget-audit2Manually: put the files in place
.github/skills/tool-budget-audit/SKILL.md1Skills CLI (Vercel)
npx skills add MiniMax-AI/mcode-stack --skill tool-budget-audit -a gemini-cli2Platform CLI
gemini skills install https://github.com/MiniMax-AI/mcode-stack --path stacks/ronny/skills/tool-budget-auditManually: put the files in place
.devin/skills/tool-budget-audit/SKILL.mdThe legacy .windsurf/skills/ path still works
Claude.ai
Zip the folder and upload it in Settings → Capabilities → Skills
Step by step, with commands for each platform: How to install skills in Claude Code: five ways
Questions about tool-budget-audit
How do I install tool-budget-audit in Claude Code?
Add the marketplace with /plugin marketplace add MiniMax-AI/mcode-stack, then install the skill with /plugin install ronny@mcode-stack.
What does tool-budget-audit do?
It measures context spent on tool schemas, call results, and model round trips. It then guides the choice between reducing the catalog, adding tool search, and using programmatic calls.
How does tool-budget-audit validate its recommendation?
It compares score, request count, and token usage before and after each change. The end-to-end experiment uses harness-experiment.
Source text
SKILL.md · MIT
--- name: tool-budget-audit description: Audit how much context an agent's tools consume, through tool definitions and tool results, and decide whether to trim the catalog, add tool search, or add programmatic tool calling. Use when an agent has many tools or MCP servers, picks the wrong tools, or burns context on tool output. --- # Tool Budget Audit Tools cost context in two ways, plus one side effect: - **Definitions.** Every tool's name, description and parameter schema sits in context whether or not it is used this turn. - **Intermediate results.** Every call puts its full output back into the conversation, and every round trip is another full model request. - **Side effect: wrong picks.** Similar names and vague descriptions make the model choose the wrong tool or pass wrong arguments. Tool search mainly saves the first cost. Programmatic tool calling (the model writes a small program that calls tools) mainly saves the second. Companion to MCode Engineering Notes 02, "Too many tools will eat your context". Plain-text checklist: `checklists/tool-budget.md`. ## 1. Measure first Work from the target project's code and real requests, not from guesses: 1. List every tool exposed to the model, grouped by source (built-in, each MCP server, plugins). 2. Estimate the token size of each tool definition with a tokenizer and sum it. Say which tokenizer you used; it is a relative measure. 3. For one or two representative tasks, record how many tool calls were made, how many model requests that caused, and how large the tool results were. 4. Note which tools are used almost every turn and which are rarely used. Report the numbers you measured. If you cannot measure something, say so instead of estimating. ## 2. Step 0 - subtract - **Merge by user intent.** Do not map APIs one to one. "List issues", "filter issues by label" and "filter by author" can be one tool with parameters. - **Make names and descriptions distinguishable.** Two tools whose descriptions read alike will be confused. State the return format. - **Group.** Put tools into namespaces by domain, keeping each group small. - **Decide what stays resident.** Read, edit, search and run-command style tools used nearly every turn should always be loaded. ## 3. Step 1 - deferred loading and tool search Consider it when there are many tools and most are rarely used. - Keep core high-frequency tools resident; make low-frequency tools searchable. - If a search miss means a tool "does not exist", check descriptions again: vague descriptions break recall. - Test tool search recall on **your own** tool set. Results on clean benchmarks do not transfer to messy real catalogs. - Tools whose catalog must be fetched over the network every turn can slow the first token. Consider keeping them resident instead. - When debugging, check what the model had actually seen at that moment, not just what the request listed. ## 4. Step 2 - programmatic tool calling Consider it when there is a lot of data to filter or aggregate, three or more dependent calls, or batch operations that could run in parallel. - Actually move tools into the program. Offering a code tool next to all the normal tools tends to go unused. - Keep at top level: tools that interact with the user, change session state, need their result read before deciding the next step (tool search, loading a skill), browser and publishing tools, and spawning subagents. - Calls from inside the program must go through the same validation, hooks, permission prompts and UI as direct calls. - Tool failures should return an error result, not throw, so one failure does not discard the whole program. - Prefer structured tool output so the program can pick the fields it needs. A program that returns raw output saves nothing. - Search the codebase for logic keyed on tool names (reminders, counters, analytics). It silently stops firing once a tool moves into the program. ## 5. Decide | Bottleneck | First move | | --- | --- | | Too many or too similar tools | Subtract, merge, rename, group | | Definitions take too much context | Tool search | | Results take too much context, too many round trips | Programmatic tool calling | | Arguments often wrong | Better descriptions and examples | The steps stack. A clean catalog makes tool search work; working tool search makes programmatic calling work. Run an end-to-end eval before and after each step and look at score, request count and tokens together (see `harness-experiment`). ## Output A short report: measured numbers, the main bottleneck, the recommended next step with concrete tool names, and what to measure afterwards.
© Ronny (MiniMax Code team) · MIT · MiniMax-AI/mcode-stack