td-builder

td-builder is an agent for Claude Code from Pantani/tdmcp. It costs 66 tokens per session (1,125 once invoked), scanned A, original, MIT.

A coding-agent role that implements one tdmcp feature from a written design specification. It creates new tool and test files without changing shared files.

In plain words
What is it for?
For adding a TouchDesigner tool: defining its inputs, connecting it to the Python bridge, registering it with the server, and testing it without a live TouchDesigner session.
Why use it?
It lets several feature implementations happen at the same time without overwriting each other's work. The required structure and offline bridge test help keep each feature consistent.

Agent for Claude Code

Install

Getting it into your agent

One page per mod, every tool's command on it. A separate URL per tool would split the same page into five that compete with each other.

agentmods
npx agentmods add agents/pantani/tdmcp/td-builder
Clone the repo
git clone --depth 1 https://github.com/Pantani/tdmcp

Made for: Claude Code.

Wrote this? Show the measurements

A badge with what this costs and how it scanned, read live from this page, so it follows the numbers instead of freezing them. Markdown for a README, HTML for a documentation site or a project page.

agentmods badge for td-builder

README.md
[![agentmods](https://agentmods.dev/badge/agents/pantani/tdmcp/td-builder.svg)](https://agentmods.dev/agents/pantani/tdmcp/td-builder)
Your own site
<a href="https://agentmods.dev/agents/pantani/tdmcp/td-builder"><img src="https://agentmods.dev/badge/agents/pantani/tdmcp/td-builder.svg" alt="Measured on agentmods" height="20"></a>
Per session 66 Only the description is in the session, so the agent can decide to use it. The body loads when it is invoked.
When invoked 1,125 The whole file, excluding the scripts and references it only reads on demand.
Security scan A 0 findings. Scan, not verified.
Origin original No closer match found in the catalogue.
Token cost

What it costs to keep this loaded

Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.

ModelPer sessionOnce invoked
Fable 5 $0.00066 $0.01125
Opus 5 $0.00033 $0.00562
Sonnet 5 $0.00013 $0.00225
Haiku 4.5 $0.00007 $0.00112

Measured 4d ago against content hash 08d6d06742cc, method: parsed. Prices are Anthropic first-party input rates as of 2026-08-30, from the pricing page.

Security

Grade A, and why

td-builder scanned grade A with 0 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 4d ago.

A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.

Nothing flagged

None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.

Origin

Copies of this mod

1 near-identical copy found in the catalogue:

.claude/agents/td-builder.md · 53 lines

How it starts

The opening of the file, as written. The whole thing — 53 lines — stays where its author put it; the contents beside it link to each section on GitHub.

td-builder — feature implementation

You implement a single tdmcp feature from a td-architect spec. Multiple builders run in parallel, so you stay strictly inside your own new files and never edit anything shared.

Skill: invoke the td-feature-build skill (via the Skill tool) at the start of your task — it holds the canonical tool-file pattern, the bridge payload recipe, the msw test pattern, and the TD gotchas that keep a build actually cooking.

Core role

  1. Create the tool file at the spec's path, exporting both …Impl(ctx, args) (pure, testable) and register…: ToolRegistrar.
  2. Define the Zod inputSchema from the spec and register with server.registerTool(name, { …, inputSchema: schema.shape }, (args) => …Impl(ctx, args)).
  3. Write the bridge work as a Python payload built with buildPayloadScript (__PAYLOAD_B64__), executed through the client and parsed back with parsePythonReport.
  4. Write one offline msw unit test in tests/unit/<feature>.test.ts that mocks the bridge — no live TouchDesigner.

Working principles — the boundary that keeps parallel builds safe

  • New files only. Create your tool file + your test file. Do not edit src/tools/layer*/index.ts, src/tools/index.ts, src/cli/agent.ts, docs, or any file another builder might also touch. Wiring is td-integrator's job. This is the single rule that lets builders run concurrently without conflicts.
  • Never throw out of a handler. Inputs are validated by the Zod schema; TD failures become friendly isError results via errorResult / runBuild / friendlyTdError. Return, don't throw.
  • ESM/NodeNext: relative imports MUST end in .js. noUncheckedIndexedAccess is on — guard array/record access.
  • Biome: 2-space indent, double quotes, semicolons, trailing commas, 100-col. Run ./node_modules/.bin/biome check <yourfiles> directly (the RTK proxy breaks npm run lint with a false parse error).
  • Don't add error handling for impossible states or speculative params beyond the spec. Match the spec's surface.
  • Honor known bridge gotchas so your build actually cooks: many params fail silently (e.g. a Level TOP has no gain — use brightness1); there are no cross-container wires (route through a Select TOP); an executePythonScript payload must assign a result variable; there is no ParMode. For GLSL TOPs: declare out vec4 fragColor;, avoid preamble #define collisions (F1/F2), and there is no built-in uTime — add your own uniform.
  • If your feature touches td/ (the Python bridge), run python3 -m py_compile on changed files and keep all TD-global usage (op, app, project) inside functions.

Read the full file on GitHub · 53 lines

Changes

What this file has done since we first saw it

Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.

  1. 4d ago First seen · 53 lines · 66 tokens per session scan A 08d6d06742cc

Subscribe to this mod's changes

td-builder is an agent published in the GitHub repository Pantani/tdmcp (39 stars, last pushed 19d ago), licensed MIT. It adds 66 tokens to every session and 1,125 once invoked, about $0.0003 per session on Opus 5. A static security scan graded it A with 0 findings. No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-30.