dsh-rea
DeepSeek Harness plugin bundle: REA (Reverse Engineer Anything) MCP tools behind a single Plugins-page switch. Ships no engine — only the connection declaration.
Install / Use
claude mcp add PandaBoby -- npx -y github:PandaBoby/dsh-reaIf the server publishes to npm under a different name, use that package instead — check the repo README.
MCP Server
Model Context Protocol server
Quality Score
Category
Development & EngineeringSupported Platforms
Tags
Our assessment of dsh-rea
dsh-rea scores 78/100 on our quality scale, 3708th of 4,705 Development & Engineering skills we index.
Its MCP Server is 4.7 KB long, well organised into 9 sections with 4 code examples: a solid amount of guidance for an agent.
It has 3 GitHub stars, so there is little community track record yet; judge it on its content.
Maintenance, license and trust
- The repository was last updated 28 days ago, so dsh-rea is actively maintained.
- It is released under the MIT license, a permissive license that allows use, modification and commercial use with attribution.
- Its trust signals score 92/100, with 1 caution from licensing, adoption, age or documentation. These come from repository metadata, not a code audit — read the skill file before letting an agent act on it.
Safety scan
No issues foundOur scan of the whole file found no instruction hijacking, hidden characters, credential access, data exfiltration or destructive commands. An AI review of the same text found nothing harmful.
AI review by kimi-k2.7-code on 2026-11-16. Automated pattern scan on 2026-11-16. It catches known dangerous patterns, not every risk — read a skill before letting an agent act on it.
dsh-rea compared with similar skills
All 4 of these similar skills score higher than dsh-rea; compare them before choosing.
| Skill | Score | Stars | Updated | Format |
|---|---|---|---|---|
| dsh-rea (this skill)by PandaBoby | 78 | 3 | 28d ago | MCP Server |
| taste-skillby Leonxlnx | 100 | 117.0k | 8d ago | CLAUDE.md |
| Agent-Reachby Panniantong | 100 | 112.4k | 9d ago | CLAUDE.md |
| rufloby ruvnet | 100 | 81.9k | today | CLAUDE.md |
| headroomby headroomlabs-ai | 100 | 79.9k | today | CLAUDE.md |
Frequently asked questions
- How do I install dsh-rea?
- Run
claude mcp add PandaBoby -- npx -y github:PandaBoby/dsh-rea. The install tabs above show the steps for each supported agent. - Which AI agents does dsh-rea work with?
- It is written for Claude Code and Claude Desktop, as a MCP Server file. Other agents that read the same format can often use it too.
- Is dsh-rea safe to use?
- Our scan of the whole file found no instruction hijacking, hidden characters, credential access, data exfiltration or destructive commands. An AI review of the same text found nothing harmful. It is MIT-licensed and scores 92/100 on trust signals. Skills are instructions an agent will follow, so read the file before installing it and do not approve commands you do not understand.
- Is dsh-rea still maintained?
- The repository was last updated 28 days ago, so dsh-rea is actively maintained.
Skill content
View source on GitHubdsh-rea
A DeepSeek Harness plugin bundle
that mounts the REA (Reverse Engineer Anything) MCP server as the rea tool
namespace — and gives it an ordinary on/off switch in the Plugins page.
This package ships no analysis engine, no tool implementation, and no model
code. It is a declaration: one MCP bridge row, plus the display metadata that
makes it a card in the plugin list. Every actual REA tool is discovered from the
rea-agents package at run time.
Why a bundle instead of a YAML snippet
Dropping the same six lines into ~/.dsh/cordis.patch.yml works, but that row is
invisible to the Plugins page — the page lists bundles (packages that carry a
dsh.bundle.patch), not free-standing rows. Wrapping the row in a bundle is what
turns it into a switch you can click, or click back off after the investigation.
The switch matters because the rea namespace is expensive while it is on: about
200 tool schemas travel in every request.
| State | Tools sent | Tool-definition context | | --- | --- | --- | | dsh-rea off | ~40 built-in | ~2K–8K tokens | | dsh-rea on | ~40 + ~207 | ~200K tokens |
Measured against rea-agents@6.5.0 on a 1M-token window: the schemas serialize to
~790 KB, which the harness prices at ~207K tokens, roughly 20% of the window. The
figure moves with the REA release; it is never zero while the bundle is on.
Install
From GitHub:
dsh plugin add github:PandaBoby/dsh-rea
Or from a local checkout, which is the form to use while developing this package:
dsh plugin add /absolute/path/to/dsh-rea
Installation also selects the bundle, so rea tools appear right away; switch it
off in the Plugins page when you are done. Grab the
matching skill from rea-agents (or from the REA repository) if you want the
model to have REA's own procedure guidance; this package deliberately does not
duplicate it.
The REA server itself is not downloaded at install time. The bridge runs
npx -y rea-agents@latest mcp, so npm fetches the newest release on the first
tool sync after the switch is turned on (and reuses its cache afterwards).
Turn it off and on
Plugins page → Installed plugins → dsh-rea — flip the switch.
- On — the MCP bridge starts, lists REA's tools, and registers them under
mcp__rea__*. - Off — the row is withdrawn, the child process is disposed, and the tool schemas stop being sent. Later requests are ~200K tokens lighter.
The switch is a profile-level setting: it applies to every session of the profile, including sessions that are already open. Turning it off while a REA call is running cancels that call.
No restart is needed when HMR is enabled (the default in Desktop and Web profiles). Without HMR, the change applies on the next start.
Configuration
Defaults live in cordis.patch.yml. Override the rea-mcp
row from your profile patch when you need to pin a REA version, raise the call
timeout for slow decompilation, or make startup failures loud:
- id: rea-mcp
config:
serverName: rea
transport: stdio
command: npx
args: ["-y", "rea-agents@6.5.0", "mcp"]
toolCallTimeoutMs: 600000
failOnStartupError: false
A patch replaces a row's whole config rather than merging into it, so restate
every key you want to keep.
| Key | Default | Meaning |
| --- | --- | --- |
| serverName | rea | Namespace of the tool names (mcp__rea__<tool>); max 32 chars |
| command / args | npx + -y rea-agents@latest mcp | The stdio server to launch |
| toolCallTimeoutMs | 60000 | Per-call budget; raise it for long analyses |
| failOnStartupError | false | true fails the plugin instead of silently exposing no tools |
| reconnect | built-in policy | Retry/backoff when the child process drops |
Verify what it costs
The context meter above the composer breaks usage into system prompt, tool
definitions, and messages. Toggle the bundle and compare the tool definitions
row — that delta is what this plugin costs. Alternatively, read the latest
request/header from a session log under ~/.dsh/sessions/ and count the
mcp__rea__* entries.
Uninstall
Plugins page → dsh-rea → Remove. Removing the bundle unregisters the row and
lets you delete the package; the npx cache is outside the profile and stays
until npm or npx clears it.
Layout
package.json name, description, icon, dsh.bundle.patch
cordis.patch.yml the one insert row: mcp-client connected to `rea mcp`
icon.svg card artwork
locale/en.json card title + description
locale/zh.json same, Chinese
Licence
MIT. REA itself is a separate project with its own licence; this package only declares how to reach it.
Related Skills
taste-skill
117.0kAgent skills that give coding agents taste: art direction, reference images and a browser check instead of generic AI pages
Agent-Reach
112.4kGive your AI agent eyes to see the entire internet. Read & search Twitter, Reddit, YouTube, GitHub, Bilibili, XiaoHongShu — one CLI, zero API fees.
ruflo
81.9k🌊 The original agent harness. Deploy intelligent multi-player swarms, coordinate autonomous workflows, and build conversational AI systems. Features adaptive memory, self-learning intelligence, federation, vector RAG integration, and native Claude Code / Codex / Hermes and many more Integrated
headroom
79.9kCompress tool outputs, logs, files, and RAG chunks before they reach the LLM. 20% fewer tokens for coding agents, 60-95% fewer tokens for JSON, same answers. Library, proxy, MCP server.
Trust signals
From repository metadata: license, adoption, age and documentation. Not a code audit — see the Safety scan above for what the skill file itself contains.
