Preface¶
In plugin-based extensibility systems like DSH, developers often need to answer the following questions before adding, replacing, upgrading, combining, or developing a capability: Does the current environment already have a usable overlay? Which projects in the ecosystem are installable, and which are just prototypes? What is the safest next step? Existing approaches typically involve manual comparison across plugin lists, code repositories, and runtime behavior, lacking a unified pre-check entry point. Below, we introduce cyanseek/dsh-landscape, a read-only capability pre-check tool for DeepSeek Harness that provides evidence-based recommendations before modifying or building capabilities.
The community directory mentioned here is an independent site and does not have an official affiliation with DeepSeek or High-Flyer.
What This Is¶
cyanseek/dsh-landscape is a read-only capability pre-check tool. Its purpose is to provide decision recommendations based on the current environment, ecosystem evidence, and constraints before modifying or building DeepSeek Harness capabilities, using natural language to describe a change request.
This plugin is maintained by cyanseek, licensed under MIT, and requires Node.js >= 22 to run. It supports usage via DSH plugins, Agent Skills, CLI, Node API, and static sites/APIs.
It differs from “directly installing plugins” or “writing code first and then verifying”: it does not execute changes but only outputs judgments and next-step recommendations. Based on verified capabilities, it explicitly does not perform operations such as installation, enabling, disabling, upgrading, uninstalling, editing profiles, or creating repositories.
Core Features¶
The main capabilities of dsh-landscape can be summarized as follows:
- Read-only capability pre-check: Use natural language to describe a change request and receive evidence-based decisions.
- No change execution: Does not install, enable, disable, upgrade, uninstall, edit profiles, or create repositories.
- Reduced upfront configuration cost: No accounts, API keys, initialization, or mandatory configurations required.
- Decision-support results: Output decision types can be
USE,INSTALL,COMPOSE,EXTEND,BUILD,WAIT,DISABLE, orINVESTIGATE. - Boundary information preserved: Maintains explicit
Do Not BuildandBuild Onlyboundaries. - Stable result contract: Includes fields such as
intent,environment,decision,risks,doNotBuild,buildOnly,nextAction, andlimitations. - Operable with insufficient runtime information: When runtime checks or live discovery is unavailable, it still uses a bundled snapshot and notes unverified content.
In terms of security and trust, verified facts also include:
- The discovery process reads attributed public metadata and does not execute code from discovered repositories.
- Absolute local paths, credential values, plugin configurations, and raw check errors are excluded from environment output.
- Archived projects are excluded from active competition by default; forks are flagged and deprioritized.
- Packages have no runtime dependencies; GitHub plugin installation does not run lifecycle build scripts.
- The
packageis not declared to be published to npm; relatednpxcommands run verified GitHub source code.
Installation and Enabling¶
First, confirm that the current environment meets Node.js >= 22.
Add a fixed version of the plugin to an existing DSH profile:
dsh plugin --profile web add github:cyanseek/dsh-landscape#2d3570aadbbd291dbfc58e2484e287bd14fa92e0
Here, --profile web is part of the DSH plugin installation command; replace it with the profile you are currently using in practice.
After installation, you can view the current configuration:
dsh --profile web --dump-config
If the plugin is no longer needed, you can remove it:
dsh plugin --profile web remove dsh-landscape
Typical Usage¶
Use Within DSH Plugins¶
After installation, you can directly ask DSH. For example:
Before adding browser automation, check this environment and the ecosystem. Tell me what to use, what not to build, and the safest next action.
This step does not command a change but asks DSH to first perform a capability pre-check: examine the current environment and ecosystem, then explain what can be used directly, what should not be duplicated, and the safest next step.
Use as an Agent Skill¶
If you want to use it temporarily without permanent installation:
npx -y skills use cyanseek/dsh-landscape --skill dsh-landscape --agent codex
If you need to install it as a portable Skill:
npx -y skills add cyanseek/dsh-landscape --skill dsh-landscape -g -a codex -y
Agent Skills are suitable for calling a pre-check before an agent modifies or builds DSH capabilities, allowing the host agent to continue with subsequent reviews.
CLI¶
You can directly pass in a request:
npx -y github:cyanseek/dsh-landscape "Should I install browser automation for DSH?"
You can also explicitly call compatible workflows:
npx -y github:cyanseek/dsh-landscape find "browser automation"
npx -y github:cyanseek/dsh-landscape analyze "Linear integration for DSH"
npx -y github:cyanseek/dsh-landscape brief "Linear integration for DSH" --format agent
npx -y github:cyanseek/dsh-landscape status --json
Note: The package is not declared to be published to npm; the above npx commands run verified GitHub source code.
Node API¶
In the Node API, you can load snapshot and alias data first, then call analyzeNeed:
import { analyzeNeed, loadAliases, loadSnapshot } from 'dsh-landscape'
const [{ snapshot }, aliasData] = await Promise.all([loadSnapshot(), loadAliases()])
const result = await analyzeNeed('Should we build browser automation?', {
snapshot,
aliasData,
})
console.log(result.decision, result.nextAction)
This code demonstrates an application integration scenario: pass in a natural language request, read the decision and nextAction fields, and use them for subsequent process judgment.
Applicable Scenarios and Notes¶
The tool is suitable for the following scenarios:
- Confirming existing coverage before adding, replacing, upgrading, combining, or developing DSH capabilities.
- Distinguishing between installable projects, prototype projects, placeholder projects, and unavailable projects.
- Providing clear
Do Not Build,Build Only, andNext Actionguidelines for subsequent agents or developers. - Calling pre-check results in automated workflows via CLI or Node API.
Note the following limitations:
- DeepSeek Harness is in developer preview and may introduce breaking changes.
- Public discovery cannot cover all GitHub, npm, private, or local projects.
- The current public runtime surface may not expose
active profile,DSH version,bundle provenance, orpeer compatibility; Landscape will report these as unknown. - Maturity judgments based on metadata are conservative and do not replace runtime acceptance testing.
- Static sites only support search; host-aware reviews require using DSH or Agent Skills.
- This plugin runs with the current
dshprocess permissions; review the source code and license before installation.
Related Links¶
- Directory page:
https://www.skillhub.cn/plugins/cyanseek/dsh-landscape - GitHub repository:
https://github.com/cyanseek/dsh-landscape