dudoxx-ai-terminal: Agent for Claude Code

.claude/agents/ddx-term-mcp-specialist.md

ddx-term-mcp-specialist is an agent for Claude Code from Dudoxx/dudoxx-ai-terminal. It costs 82 tokens per session (869 once invoked), scanned A, original, MIT.

A coding assistant for the ddx-term-mcp package, which lets an agent communicate with a shared tmux terminal session. It covers the server, terminal commands, screen reading, and publishing setup.

In plain words
What is it for?
Use it when changing files under ddx-term-mcp/src or its publishing configuration, including terminal sessions, command handling, screen snapshots, and the bundled release.
Why use it?
It keeps terminal-related changes in one place and preserves the package’s rules, such as using a simple command runner instead of a pseudo-terminal.

Agent for Claude Code

Written for Claude Code: paths in frontmatter. Also seen: model in frontmatter; mentions CLAUDE.md.

This is Dudoxx/dudoxx-ai-terminal's own configuration. It tells Claude Code how to work on dudoxx-ai-terminal itself, so it is not a mod to install elsewhere. Copy it as a starting point and replace the rules that are about this project. Everything dudoxx-ai-terminal configures →

Reuse

Borrowing it

Nothing to install: this file belongs to Dudoxx/dudoxx-ai-terminal. Take a copy, put it at the same path in your own repository, and replace the rules that are about this project with yours.

Copy the file
curl -O https://raw.githubusercontent.com/Dudoxx/dudoxx-ai-terminal/main/.claude/agents/ddx-term-mcp-specialist.md
Clone the repo
git clone --depth 1 https://github.com/Dudoxx/dudoxx-ai-terminal

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 ddx-term-mcp-specialist

README.md
[![agentmods](https://agentmods.dev/badge/agents/dudoxx/dudoxx-ai-terminal/ddx-term-mcp-specialist/github.svg)](https://agentmods.dev/agents/dudoxx/dudoxx-ai-terminal/ddx-term-mcp-specialist)
Your own site
<a href="https://agentmods.dev/agents/dudoxx/dudoxx-ai-terminal/ddx-term-mcp-specialist"><img src="https://agentmods.dev/badge/agents/dudoxx/dudoxx-ai-terminal/ddx-term-mcp-specialist/github.svg" alt="Measured on agentmods" height="20"></a>

Or the 80×15 button, for a site that already has a row of RSS and ATOM ones. Only the verdict fits; the numbers stay here.

agentmods 80×15 button for ddx-term-mcp-specialist

Your own site · 80×15
<a href="https://agentmods.dev/agents/dudoxx/dudoxx-ai-terminal/ddx-term-mcp-specialist"><img src="https://agentmods.dev/badge/agents/dudoxx/dudoxx-ai-terminal/ddx-term-mcp-specialist.svg" alt="Reviewed on agentmods" width="80" height="20"></a>
Per session 82 Only the description is in the session, so the agent can decide to use it. The body loads when it is invoked.
When invoked 869 The whole file, excluding the scripts and references it only reads on demand.
Security scan A 1 finding. A grade says what 26 rules found in the file — not that it is safe.
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 5.1 $0.00082 $0.00869
Opus 5 $0.00041 $0.00434
Sonnet 5 $0.00016 $0.00174
Haiku 4.5 $0.00008 $0.00087

Measured 12d ago against content hash 381d9a9d8944, method: parsed. Prices are Anthropic first-party input rates as of 2026-09-12, from the pricing page.

Security

Grade A, and why

ddx-term-mcp-specialist scanned grade A with 1 finding 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 12d 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.

Runs shell commandslowCapability

Expected in a hook, worth knowing in a rule or an instructions file.

- **NO PTY** — no `node-pty`, no `pty.spawn`, no raw-shell `child_process`. tmux only.
.claude/agents/ddx-term-mcp-specialist.md · 52 lines

How it starts

The opening of the file, as written. The whole thing — 52 lines — stays where its author put it; the contents beside it link to each section on GitHub.

You are the @dudoxx/ddx-term-mcp specialist — the MCP stdio (JSON-RPC 2.0) server that lets Claude drive the SAME shared tmux session a human is watching.

Cold Start (MANDATORY)

You start with ZERO context from the parent. At the start of EVERY task:

  1. Read context/agents/ddx-term-mcp-specialist_memory.md if it exists.
  2. Read context/agents/ddx-term-mcp-specialist_extra_learned_instructions.md if it exists.
  3. Read ddx-term-mcp/CLAUDE.md and context/_invariants.md — the binding contract.
  4. Read ONLY the task's named files.
  5. If a referenced path/fact is unverifiable from disk, ask — do not fabricate.

Ownership

  • src/server.ts — stdio wiring (entry → dist/server.js); ListTools + CallTool handlers.
  • src/tools/term-*.tool.ts + registry.ts — the 10 verbs (create/list/destroy, send/read/ wait-for/signal/ps, panes/snapshot).
  • src/tmux/tmux.client.ts — the ONLY thing that shells out to tmux (execFile).
  • src/{read-cursor,terminal-map,registry-resolver,context,errors}.ts.
  • tsup.config.ts — the publish bundle (inlines @ddx/term-contract via noExternal).

Load-bearing invariants (NEVER violate)

  • NO PTY — no node-pty, no pty.spawn, no raw-shell child_process. tmux only. no-pty.spec.ts greps the source and MUST stay green.
  • term_send: send-keys -l <text> (LITERAL) + a SEPARATE Enter event. NEVER \n. Control keys → term_signal, never term_send.
  • term_snapshot = visible viewport (capture-pane -p WITHOUT -S). term_read = scrollback delta via per-terminalId read-cursor, capped by DDX_TERM_MAX_READ_LINES.
  • Address by terminalId; term_signal(pid) validates pid ∈ the terminal's process tree.
  • Never exceed DDX_TERM_MAX_TERMINALS; never target a session ≠ $DDX_TERM_SESSION.
  • All input schemas from @ddx/term-contract (TERM_TOOL_INPUT_SCHEMAS). Zero any.

Publish (the corrected reality)

The contract is BUNDLED into dist/server.js by build:bundle (tsup, noExternal), wired as prepublishOnly. The package IS publish-ready to public npm. The contract stays private + in changeset ignore[] by design. The README's "npx not available yet" note is STALE.

Read the full file on GitHub · 52 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. 12d ago First seen · 52 lines · 82 tokens per session scan A 381d9a9d8944

Subscribe to this mod's changes

ddx-term-mcp-specialist is an agent published in the GitHub repository Dudoxx/dudoxx-ai-terminal (1 stars, last pushed 2mo ago), licensed MIT. It adds 82 tokens to every session and 869 once invoked, about $0.0004 per session on Opus 5. A static security scan graded it A with 1 finding (runs shell commands). No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-31.

Related

Other agents, from other repositories