npm package discovery and stats viewer.

Discover Tips

  • General search

    [free text search, go nuts!]

  • Package details

    pkg:[package-name]

  • User packages

    @[username]

Sponsor

Optimize Toolset

I’ve always been into building performant and accessible sites, but lately I’ve been taking it extremely seriously. So much so that I’ve been building a tool to help me optimize and monitor the sites that I build to make sure that I’m making an attempt to offer the best experience to those who visit them. If you’re into performant, accessible and SEO friendly sites, you might like it too! You can check it out at Optimize Toolset.

About

Hi, 👋, I’m Ryan Hefner  and I built this site for me, and you! The goal of this site was to provide an easy way for me to check the stats on my npm packages, both for prioritizing issues and updates, and to give me a little kick in the pants to keep up on stuff.

As I was building it, I realized that I was actually using the tool to build the tool, and figured I might as well put this out there and hopefully others will find it to be a fast and useful way to search and browse npm packages as I have.

If you’re interested in other things I’m working on, follow me on Twitter or check out the open source projects I’ve been publishing on GitHub.

I am also working on a Twitter bot for this site to tweet the most popular, newest, random packages from npm. Please follow that account now and it will start sending out packages soon–ish.

Open Software & Tools

This site wouldn’t be possible without the immense generosity and tireless efforts from the people who make contributions to the world and share their work via open source initiatives. Thank you 🙏

© 2026 – Pkg Stats / Ryan Hefner

@deepwatch/dsh-bundle

v0.1.4

Published

DeepWatch's capabilities, installable into an existing DeepSeek Harness profile

Readme

@deepwatch/dsh-bundle

DeepWatch's capabilities, installable into a DeepSeek Harness profile you already run. It gives that agent senses, and an independent answer to did that actually work? — without asking you to move to a different product.

The full application is DeepWatch; this package is the part of it that composes into somebody else's Harness.

Install

dsh plugin --profile web add @deepwatch/dsh-bundle

Name your own profile in place of web if you have one, and boot the same one:

cd <your-project>
dsh --profile web

There is no web subcommand — dsh web is dsh --profile web, so writing dsh --profile <name> web boots <name> and hands web to the app as an argument. And start it from the directory you are working in: the receipt journal is written under the launch directory, so a Host started elsewhere journals elsewhere and the Library reports empty.

That is the whole installation. The package declares dsh.bundle.patch, so DSH reconciles it into the profile's layer stack and applies cordis.patch.yml after its own layers.

This bundle's media capabilities are the engine's, so install it with the extras rather than bare:

pip install 'watch-skill[standard,ocr]'

[standard] is frames, retrieval and MCP; [ocr] reads on-screen text. A bare pip install watch-skill leaves you with a Core that cannot extract a frame — the Bridge connects to it and every media tool answers perceive.missing_dependency on the first video.

watch-skill is on PyPI. This bundle is built against 1.4.3 and needs at least that version: on every engine before it, watch_moment raises rather than answering, because the Bridge tried to iterate its result instead of serialising it. The Bridge finds the executable on PATH and connects on its own; watch-skill doctor names the exact command for anything still missing.

Stability

0.1.4 — a stable release. Stable means tested, documented and supported — not 1.0. This is a pre-1.0 line, and semantic versioning gives 0.x no compatibility guarantee across minor versions: a 0.MINOR bump may change or remove surface, and a patch will not. Depend on it with a tilde range (~0.1.4) if you want that difference enforced by your lockfile rather than by a changelog. The usual major-version promise starts at 1.0.

What you get

| Tool | What it does | |---|---| | watch_capabilities | what Watch can actually do here, and what is missing | | watch_list_sources | the sources Watch has indexed | | watch_ask_source | ask a question about a source, answered with timestamped evidence | | watch_get_evidence | resolve a citation and check whether it is still current | | watch_verify | run a verification contract and return an independent verdict |

Plus a browser half that renders verification results as themselves — UNVERIFIED looks like UNVERIFIED, and green is reserved for VERIFIED — and a system-prompt section telling the agent that a tool returning without error is not proof that anything worked.

Without Watch Core installed

Watch reports itself unavailable. auto does not substitute a mock: the Bridge says core_missing, every Watch capability reads Unavailable, and the install step is attached to each of them. Nothing pretends to work, and the rest of your Workspace is unaffected.

If Watch Core is installed but fails to start, that is reported as a fault too, with its own fix. Neither case is ever answered from the mock — a Bridge that did that would give you a green Workspace and no indication that nothing you ask will ever be answered.

Configuration

| Key | Default | Meaning | |---|---|---| | transport | auto | auto connects the engine, and reports it unavailable when it is absent or broken — it never selects the mock. stdio is the same transport named explicitly. mock is the test-only in-process backend and must be asked for by name. | | command | watch-skill | the executable that starts Watch Core in Bridge mode | | args | [bridge] | its arguments | | startupTimeoutMs | 45000 | how long the engine has to start. Generous on purpose: a first spawn against a freshly created virtualenv pays for a cold import that a 10s backstop reported as a dead engine. | | requestTimeoutMs | 30000 | deadline for a request that carries none | | dataDir | (empty) | where this profile's engine keeps its Library, Memory, receipts and indexes. Empty leaves the engine its own ~/.watch-skill, which two profiles on one machine would then share. deepwatch setup composes a directory inside the profile instead. It is a default: an exported WATCHSKILL_DATA_DIR still wins, and nothing is migrated into it. | | autoConnect | true | connect during activation rather than on first use |

A deployment that requires the engine should pin transport: stdio, which never falls back and so fails loudly when it is missing. A patch replaces the targeted row's whole config, so restate every key you want to keep.

Uninstall

dsh plugin --profile web remove @deepwatch/dsh-bundle

The rows are purely additive, so removing the bundle leaves the host profile exactly as it was.

Attribution

Built on DeepSeek Harness · Powered by Watch Skill. DeepWatch and Watch Skill are independent projects and are not affiliated with or endorsed by DeepSeek.