A native DSH diagnostic plugin that surfaces redundant context, rank-shadow conflicts, and trimming priorities.
DSH integration
Native runtime
Author-claimed
Safety audit
Unaudited
Last verified
2026-09-08
License
BSD-3-Clause
01What can it help you accomplish?
Measure and diagnose the context automatically injected into DSH requests
Token estimates, duplicate and rank-shadow findings, plus prioritized trimming suggestions
DSH users who need to reduce instruction, skill, tool-schema, and MCP context overhead
02How to install into DeepSeek Harness
Prerequisites
- DSH `>= 0.1.2-rc.1`
Installation steps
- 01
Install with `dsh plugin --profile web add "github:Zhenyu98/dsh-context-doctor#main"`.
$ dsh plugin --profile web add "github:Zhenyu98/dsh-context-doctor#main"
- 02
Restart DSH web and invoke `context_audit` in a new session.
Verify the integration
- Run `dsh --profile web --dump-config | grep context-doctor` and confirm the plugin is inserted.
03DSH integration and capability boundaries
Native DSH web-profile plugin with a web panel and `context_audit` model tool
Context-audit panel and tool
DSH instruction chains, skill catalog entries, tool schemas, and MCP tool names→A sectioned audit report with token estimates, conflict findings, and suggestions
Reads files and metadata only; files larger than 256 KB are skipped
04Who is it for? When not to use it?
Good for
- DSH users who need to reduce instruction, skill, tool-schema, and MCP context overhead
Not for
- Token counts are heuristic estimates for comparison, not the model tokenizer’s exact values.
05Compatibility, maintenance and safety notes
- Token counts are heuristic estimates for comparison, not the model tokenizer’s exact values.
Author README tracked from the referenced commit
06Frequently asked questions
What does `context_audit` inspect?
It reports DSH instruction chains, skill catalog entries, tool schemas, MCP tools, conflicts, and suggestions.
Will it edit my project files?
No. The documented boundary is read/stat/list only, with no writes, deletes, or execution of audited objects.
Why can its number differ from the DSH meter?
It uses a heuristic token estimate for comparison; the model tokenizer remains the source for exact token counts.
07Related DSH workflows
dsh-context
by bowenliang123
Context dashboard tab + /context command showing context-window composition, per-turn history, compaction/prune events, per-message token costs
api-relay-audit
by toby-bridges
Local security audit for AI API relays and LLM proxies: detects prompt injection, model substitution, tool-call rewriting, SSE anomalies, error leakage, and Web3 wallet risks.
jingyun-dsh
by jingyunstudio
A one-stop AI commercialization desktop client built on Jingyun Studio plus DeepSeek Harness (DSH).
awesome-deepseek-harness-plugins
by zhiyuan-fan
Curated DeepSeek Harness (DSH) plugins, extensions, tools, skills, clients, runtimes, integrations, and verified references — English and Chinese.
08Data and sources
DSH 上下文注入审计插件：看清模型每个请求到底背着多少上下文，找出重复、冲突与浪费 token 的注入物。
This page is generated from the project’s public documentation, repository metadata and a structured parse of DSH Plugins; last verified on 2026-09-08. Found an error? Submit a correction.
Best DeepSeek Harness Plugins
Twelve plugins worth installing first — picked from the whole catalog, across every category.