What is deepseek-subagent-mcp CLAUDE.md?

A set of Claude Code instructions for an MCP server that lets other MCP clients delegate work to DeepSeek Harness, an open-source agent runtime. It also supervises the delegated agent and returns a checked result with a concise handoff.

Use it to start, wait for, continue, list, cancel, and inspect delegated agent runs. Local tests cover the server and its safeguards; live runs require a DeepSeek API key and use paid API tokens.

It helps separate delegated work from the main agent while controlling tool calls and managing the child process. It provides a completion result instead of requiring the client to interpret a full raw transcript.

How gaztrabisme/deepseek-subagent-mcp configures Claude Code

Reuse

Borrowing it

Nothing to install: this file belongs to gaztrabisme/deepseek-subagent-mcp. 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/gaztrabisme/deepseek-subagent-mcp/main/CLAUDE.md

Clone the repo

git clone --depth 1 https://github.com/gaztrabisme/deepseek-subagent-mcp

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.

README.md

[![agentmods](https://agentmods.dev/badge/instructions/gaztrabisme/deepseek-subagent-mcp/claude-md/github.svg)](https://agentmods.dev/instructions/gaztrabisme/deepseek-subagent-mcp/claude-md)

Your own site

<a href="https://agentmods.dev/instructions/gaztrabisme/deepseek-subagent-mcp/claude-md"><img src="https://agentmods.dev/badge/instructions/gaztrabisme/deepseek-subagent-mcp/claude-md/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.

Your own site · 80×15

<a href="https://agentmods.dev/instructions/gaztrabisme/deepseek-subagent-mcp/claude-md"><img src="https://agentmods.dev/badge/instructions/gaztrabisme/deepseek-subagent-mcp/claude-md.svg" alt="Reviewed on agentmods" width="80" height="20"></a>

Per session 3,189 This file is loaded in full into every session.

When invoked 3,189 The same file — it is already loaded in full.

Security scan C 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.

Model	Per session	Once invoked
Fable 5.1	$0.03189	$0.03189
Opus 5	$0.01595	$0.01595
Sonnet 5	$0.00638	$0.00638
Haiku 4.5	$0.00319	$0.00319

Measured 9d ago against content hash cd138a0a0bed, method: parsed. Prices are Anthropic first-party input rates as of 2026-09-10, from the pricing page.

Security

Grade C, and why

deepseek-subagent-mcp CLAUDE.md scanned grade C 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 9d 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.

Reaches for credential fileshighPrivilege escalation

SSH keys, cloud credentials, git-credentials, .npmrc, /etc/shadow: reading these is how a config file becomes a credential leak.

4. **A read-only verb is not a read-only call.** `cat ~/.ssh/id_rsa` is `cat`. Sensitive paths are

CLAUDE.md · 218 lines

How it starts

The opening of the file, as written. The whole thing — 218 lines — stays where its author put it; the contents beside it link to each section on GitHub.

CLAUDE.md

This file provides guidance to Claude Code (claude.ai/code) when working with code in this repository.

What this is

An MCP server that exposes DeepSeek Harness (dsh, DeepSeek's open-source agent runtime, released 2026-08-13, MIT) as a delegatable subagent to Claude Code, Codex, or any other MCP client. Six tools: dsh_delegate, dsh_await, dsh_continue, dsh_list, dsh_cancel, dsh_transcript.

It ships four things upstream does not: a hardened plugin composition, an external supervision layer that gates the child's tool calls, process-lifetime management the wire protocol cannot provide, and a result contract — verified completion and a distilled handoff rather than a raw transcript.

Commands

uv sync
uv run pytest                                     # 153 unit tests, no API key, no network
uv run ruff check .
uv run pytest tests/test_guard.py -q              # the classifier alone
uv run pytest tests/test_runs.py::test_run_deadline_kills_the_agent
uv build
uv run deepseek-subagent-mcp                      # stdio; a client drives it

Live smokes cost real tokens, need DEEPSEEK_API_KEY, and are not collected by pytest. Run them by path after touching runs.py, server.py, guard.py, verify.py, supervisor.py, or the YAML:

DEEPSEEK_API_KEY=sk-... uv run python tests/smoke_mcp.py        # all 6 tools over stdio
DEEPSEEK_API_KEY=sk-... uv run python tests/smoke_task.py       # real file writes, 2-turn session
DEEPSEEK_API_KEY=sk-... uv run python tests/smoke_result.py     # distillation, cap, run archive
DEEPSEEK_API_KEY=sk-... uv run python tests/smoke_limits.py     # reaper + run deadline
DEEPSEEK_API_KEY=sk-... uv run python tests/smoke_supervisor.py # allow/deny through the hook
DEEPSEEK_API_KEY=sk-... uv run python tests/smoke_escalation.py # sampling + elicitation tiers

smoke_supervisor.py is the one that proves the security story. smoke_task.py proves the product works. smoke_result.py proves the caller gets a result rather than a transcript.

Read the full file on GitHub · 218 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.

9d ago First seen · 218 lines · 3,189 tokens per session scan C cd138a0a0bed

Subscribe to this mod's changes

deepseek-subagent-mcp CLAUDE.md is an instructions file published in the GitHub repository gaztrabisme/deepseek-subagent-mcp (0 stars, last pushed 7d ago), licensed MIT. It adds 3,189 tokens to every session, about $0.0159 per session on Opus 5. A static security scan graded it C with 1 finding (reaches for credential files). No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-31.

Building agents

next.js AGENTS.md

AGENTS.md instructions for vercel/next.js, covering next.js development guide, codebase structure, monorepo overview, core package: packages/next and other important packages.

vercel/next.js · 7,296 tokens

codex AGENTS.md

AGENTS.md instructions for openai/codex, covering rust/codex-rs, the codex-core crate, code review rules, crate api surface and model visible context.

openai/codex · 5,153 tokens

vscode buildNext.instructions.md

Working notes and architecture documentation for the new esbuild-based build system in build/next. Use when making changes to the new build pipeline (transpile/bundle commands, NLS plugin, source-map handling, resource copying, or self-hosting watch tasks).

microsoft/vscode · 6,785 tokens

deepseek-subagent-mcp: Instructions file for Claude Code