agentmods.dev

dstools CLAUDE.md

dstools CLAUDE.md is an instructions file for coding agents from lijiatuk/dstools. It costs 1,053 tokens per session, scanned A, original, MIT.

A project guide for dstools, a local MCP server that adds image understanding and deep research to the DeepSeek-V4 model. It documents the server's commands, design, model settings, and how to add a tool.

In plain words
What is it for?
Use it when installing dstools, configuring its models and API keys, learning how its tools work, or extending the server with another tool.
Why use it?
It gives developers one reference for understanding and modifying the repository. It explains the project-specific setup instead of leaving contributors to infer the architecture from the code.

Instructions file

not rated 1 2mo ago A scan 1,053 tokens original MIT
View the source file lijiatuk/dstools
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 instructions/lijiatuk/dstools/claude-md
Clone the repo
git clone --depth 1 https://github.com/lijiatuk/dstools

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 dstools CLAUDE.md

README.md
[![agentmods](https://agentmods.dev/badge/instructions/lijiatuk/dstools/claude-md.svg)](https://agentmods.dev/instructions/lijiatuk/dstools/claude-md)
Your own site
<a href="https://agentmods.dev/instructions/lijiatuk/dstools/claude-md"><img src="https://agentmods.dev/badge/instructions/lijiatuk/dstools/claude-md.svg" alt="Measured on agentmods" height="20"></a>
Per session 1,053 This file is loaded in full into every session.
When invoked 1,053 The same file — it is already loaded in full.
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 $0.01053 $0.01053
Opus 5 $0.00526 $0.00526
Sonnet 5 $0.00211 $0.00211
Haiku 4.5 $0.00105 $0.00105

Measured 4d ago against content hash 0c7d52f29e51, method: parsed. Prices are Anthropic first-party input rates as of 2026-08-30, from the pricing page.

Security

Grade A, and why

dstools CLAUDE.md 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.

CLAUDE.md · 76 lines

How it starts

The opening of the file, as written. The whole thing — 76 lines — stays where its author put it; the contents beside it link to each section on GitHub.

CLAUDE.md — dstools

A reference for anyone (human or AI) working in this repo.

What this is

dstools is an MCP server that gives DeepSeek-V4 (a text-only model) two capabilities it lacks natively: image understanding and deep research. It exposes 5 tools over stdio / Streamable HTTP / SSE.

DeepSeek-V4 specifics (verified 2026-06-30 via the official API docs)

  • API is OpenAI-compatible; base_url=https://api.deepseek.com (Anthropic format at /anthropic). Use the openai SDK.
  • Models: deepseek-v4-flash (cheap/fast) and deepseek-v4-pro (heavy). The legacy deepseek-chat / deepseek-reasoner are deprecated 2026-07-24.
  • Thinking mode: toggle via extra_body={"thinking": {"type": "enabled"|"disabled"}} (default enabled); effort via reasoning_effort (high/max). In thinking mode temperature/top_p are silently ignored; CoT is returned as message.reasoning_content.
  • JSON output: response_format={"type": "json_object"}. Tool calls: standard OpenAI function format. Context caching: automatic (keep system prompts constant to benefit).

Adaptations live in src/dstools/llm/deepseek.py: per-call thinking toggle, reasoning_content capture, JSON mode, retry with backoff, pro/flash model split (pro for synthesis, flash for cheap planning).

Commands

uv sync --extra dev          # install
uv run dstools serve         # run MCP server (stdio)
uv run dstools serve --transport http --port 8000
uv run dstools inspect       # list tools + param schemas
uv run dstools doctor        # what's configured (keys/providers)
make lint && make typecheck && make test
uv run python examples/mcp_client_demo.py   # end-to-end stdio smoke test

Architecture

  • server.py — FastMCP instance + tool registration (create_server()).
  • cli.py — dstools CLI (serve / inspect / doctor / cache / version).
  • config.py — Settings (pydantic-settings, env + .env). DeepSeek fields also accept the generic LLM_* aliases (LLM_API_KEY/LLM_BASE_URL/LLM_MODEL/...).
  • runtime.py — lazy singletons; wraps search/fetcher in the cache when enabled.
  • cache.py — opt-in on-disk retrieval cache (RetrievalCache + caching wrappers).
  • llm/ — DeepSeekClient (V4-aware) + VisionClient (pluggable multimodal).
  • search/ — providers: DuckDuckGoSearchProvider (keyless default, ad-filtered, retry/backoff) + BraveSearchProvider + TavilySearchProvider (both keyed).
  • web/fetcher.py — async page fetch (streamed, size-capped) + HTML→Markdown; defines the Fetcher Protocol satisfied by PageFetcher and the cache wrapper.
  • tools/ — logic functions (testable, DI) + register(mcp) wrappers; deep_research = plan → round loop (search/fetch/refine) → rerank → synthesize. _ctx.py makes ctx calls defensive (never crash on missing session).
  • utils/ — image I/O + text helpers.

Read the full file on GitHub · 76 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 · 76 lines · 1,053 tokens per session scan A 0c7d52f29e51

Subscribe to this mod's changes

dstools CLAUDE.md is an instructions file published in the GitHub repository lijiatuk/dstools (1 stars, last pushed 2mo ago), licensed MIT. It adds 1,053 tokens to every session, about $0.0053 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 2026-08-31.

Data and AIResearch
Related

Other instructions, from other repositories

GPT-RAG release.instructions.md

Instructions for Azure/GPT-RAG, a project described as: Enterprise-grade accelerator for agentic RAG on Azure. Built on Microsoft Foundry with Foundry IQ as the default retrieval backend, Microsoft Agent Framework orchestration, Zero-Trust architecture and IaC.

Azure/GPT-RAG · 207 tokens

apex-accelerator vendor-prompting.instructions.md

Vendor prompting best-practice rules for Anthropic Claude and OpenAI GPT-5.6-Terra agents and prompts. Each rule cites a rule ID in the vendor-prompting skill rules.json registry. Validator: npm run lint:vendor-prompting.

jonathan-vella/apex-accelerator · 1,488 tokens

ken CLAUDE.md

Claude Code instructions for townsendmerino/ken, covering claude.md, what this is, repository ownership (read this first), commands and embedding parity & golden fixtures (now in aikit).

townsendmerino/ken · 6,633 tokens

rag-code-mcp copilot-instructions.md

Instructions for doITmagic/rag-code-mcp, covering copilot instructions - ragcode mcp, ⚖️ the golden rule, project overview, architecture & patterns and developer workflows.

doITmagic/rag-code-mcp · 568 tokens

ZipAI CLAUDE.md

Claude Code instructions for nickdesi/ZipAI, covering claude.md — zipai: ultra-dense token optimizer, rules, 1. zero filler, 2. ambiguity and 3. prompt caching.

nickdesi/ZipAI · 1,204 tokens

gpu-ai-skills CLAUDE.md

Claude Code instructions for intel/gpu-ai-skills, covering claude.md, what this repository is, repository structure, commands and validation (required before any skill change).

intel/gpu-ai-skills · 1,541 tokens