DeepSeek Harness 插件

dsh-infra-observability

DSH-Plugin: structural observability layer — real tool/skill usage recording (tools/result), skill-catalog audit, and a watchdog with health/error events. No model self-reporting.（英文原文）

插件市场与管理

跳到安装方式

GitHub 仓库: Bryan-cmf/dsh-infra-observability
最近更新: 2026年8月17日
分类: 插件市场与管理
GitHub stars: 0
载体类型: plugin
目录来源: dshworks/awesome-dsh-plugins
目录证据: 上游声明已找到 dsh.bundle
证据路径: package.json#dsh.bundle
核对版本: 0.1.0-rc.8
上游核对日期: 2026-08-20

该证据由上游目录提供。本站没有安装、运行或安全审核这个插件。

安装

默认先复制一段 Prompt，让 Agent 读 GitHub 仓库和源码；需要自己装时再切到命令。

复制这段 Prompt，发给 DSH、Codex 或其他 Agent，让它先读 GitHub 仓库和源码。

请先不要安装或执行任何命令。阅读这个插件的 GitHub 仓库、README 和关键源码，然后用清楚、直接的方式回答以下问题，帮助我判断它是否适合我的需求：

1. 这个插件是什么，解决什么问题；
2. 适合哪些用户和典型使用场景；
3. 安装后如何使用，并给出一个最小使用示例；
4. 有哪些已知限制，以及隐私、安全、兼容性或维护风险；
5. 给出“推荐 / 有条件推荐 / 不推荐”的明确建议和理由。

请区分仓库明确说明、根据源码推断和未知信息。证据不足时请明确说明，不要猜测或照抄 README。

GitHub：https://github.com/Bryan-cmf/dsh-infra-observability
插件名：dsh-infra-observability
作者：Bryan-cmf

检查来源文件

安装前先看这个插件目录里的 README 和其他文件。

文件资源管理器3 个文件

README.md来源说明 · 只读预览

> ⚠️ 此套件已併入 @bryan-cmf/dsh-insights(2026-08-17)。此 repo 僅保留作合併前歷史,後續迭代請到 dsh-insights。

🔭 Infra Observability — DSH-Plugin

> 結構化觀測層:真實的工具/技能使用記錄(tools/result)、技能目錄審計、看門狗。零模型自報——平台記什麼,報告就寫什麼。

![npm](https://www.npmjs.com/package/@bryan-cmf/dsh-infra-observability) ![License: MIT](LICENSE) ![DSH](https://github.com/deepseek-ai/deepseek-harness)

A DeepSeek Harness host plugin that replaces model self-reporting with structural recording:

tools/result listener — every real tool execution (name + outcome) is

recorded by the platform into a ring buffer and per-day aggregates.

agent/error / agent/status — error totals and live-agent tracking.
Watchdog — timer-driven alert when agent errors spike (≥ N in 5 min).
Three tools for the model and humans:

Tool	What it reports
`usage_report`	Per-day tool/skill aggregates (calls / ok / err) + recent executions
`audit_skills`	Catalog totals, provider spread, warnings (missing description, non-kebab names), today's skill usage
`infra_health`	Uptime, records kept, error counts, running agents, catalog size, watchdog state

Dashboard (browser)

The client half registers an 「觀測」view tab (order 20, right of chat / trajectory) rendering a live per-session dashboard from the infraView session projection: turn counts, failure count, tool/skill TOP lists with success-rate bars, and the recent execution trail. Data updates as events commit — no refresh needed.

Install

Path A — Host composition (recommended, always-on)

Add to the profile's package.json dsh.profile.bundles (the package ships a dsh.bundle.patch that inserts the plugin row):

"dsh": { "profile": { "bundles": ["@bryan-cmf/dsh-infra-observability"] } }

Path B — Agent preset row

- id: infra-observability
  name: '@bryan-cmf/dsh-infra-observability'

Requirements

DSH >= 0.1.0-rc.6 (services: tools, timer, skills — all part of the

shipped dsh-base bundle)

Configuration

Key	Default	Meaning
`maxRecords`	`5000`	Ring-buffer cap for recent execution records
`healthIntervalMs`	`60000`	Watchdog check interval
`errorAlertThreshold`	`5`	Alert at ≥ this many agent errors per 5-minute window

Scope notes

Aggregates are process-lifetime (like dsh-skill-trail); the durable source

of truth remains the session log. Cross-restart persistence is planned.

Companion package: @bryan-cmf/dsh-skill-trail

renders the per-reply badge row; this package is the backend layer.

Development

pnpm install --store-dir .pnpm-store
pnpm build

License

MIT

返回搜索打开 GitHub 仓库