Should you try it?
Try it in a disposable DSH Desktop Beta profile if per-turn skill visibility is the problem you need to solve. The documented management controls are relevant to that workflow, but our sandbox build does not establish that the panel works in your desktop installation.
The README describes dsh-echocat-skill-panel 5.1.3 as a DSH Desktop Beta plugin for tracking per-turn skill use and installing, removing, and updating skills. It says the plugin derives skill use from conversation events, including automatic loads and manual slash invocations. Those are documented capabilities, rather than observations from our desktop testing.
What our sandbox established
Our run on October 5, 2026 tested commit 1f98da6 in our sandbox; it did not verify the plugin inside a running DSH Desktop application.
Installation recorded exit 0 in 5 seconds, the build recorded exit 0 in 6 seconds, and the test command recorded exit 1 in 4 seconds. Our harness stores each step's outcome, exit code, duration, timeout flag, and a log tail. Reading those fields together matters here: the installation tail contains an earlier npm ci lockfile error even though the enclosing step succeeded.
The build tail confirms that the build script wrote lib/client.js. This verifies bundle generation in that run. It does not verify desktop loading, a conversation workload, skill installation, update recovery, or interface behavior.
The stored sandbox description names the lab-node:22 image, while the test tail reports Node.js v20.20.2 and runtime detection records Node major 20. The image label alone does not establish which Node executable ran the command.
The MrKeyoor testing methodology explains how command outcomes and measurement limits are reported. Its supplied October 31 snapshot provides provenance context; it does not establish that the October 5 run used that later procedure unchanged.
Why the test command failed
The raw failure names @deepseek-ai/cordis as the package that test/smoke.mjs could not import. The tail also contains a successful tracker/store footer before the missing-package error. A successful subset footer and a failed overall command can coexist; this evidence does not establish an internally inconsistent release gate.
We do not report that footer as an aggregate passing-test count. The attached aggregate tests field is null, and exhaustive unique coverage was not verified. There was no new execution or complete-log reconciliation. The supported conclusion is narrower: the test command could not complete successfully in our sandbox because the smoke test encountered a missing package. Whether a properly configured DSH host resolves that import remains untested here.
The stored npm audit result contains 0 advisories. Scanned-package coverage is unknown because the record has no scannedPackages field. This is an advisory result, not a security assessment of the plugin or the skills it installs.
What the management controls change
According to the README, installation accepts repository and subdirectory addresses, SKILL.md links, ZIP links, uploaded Markdown or ZIP files, and pasted skill content. Same-named skills are backed up before replacement, and deletion moves files into backup rather than permanently removing them.
The README says source information is stored in .echocat.json beside an installed skill, including its address, branch, subdirectory, installation commit, and content fingerprint. Update checks use git ls-remote. Updates move the old copy to backup, while the confirmation warns when local edits will be overwritten.
Disabling a skill moves its directory out of the location DSH watches. That gives users a concrete filesystem operation to inspect. We did not exercise these operations, verify restoration, or review installed skill content for safety.
Discovery tools complement the panel
Awesome DeepSeek Harness is a curated directory of plugins, skills, MCP servers, patch/profile layers, orchestrators, and interfaces for DeepSeek Harness. Use it to find projects to inspect; its supplied description establishes a discovery role rather than the panel's per-turn reporting role.
Awesome DSH Plugins publishes a machine-readable registry whose source of truth is data/plugins.json. Choose it when you need registry data or discovery integration. Both directories can complement a panel installation; neither supplied extract establishes a replacement for this plugin's conversation-event reporting.
How to read the scores
These editorial ratings use a five-point rubric: 1 means substantial documented obstacles, 2 means significant limitations, 3 means useful evidence with material gaps, 4 means strong evidence for the stated audience, and 5 requires broad verification with few unresolved limits. They are not benchmark results or adoption measurements.
Setup receives 3 because the recorded install and build succeeded, but desktop wiring and restart remain outside our verification. Documentation receives 4 for the Chinese-reading audience because the supplied README explains installation, restart boundaries, source tracking, backup behavior, and development checks. Community receives 3 as a limited assessment of public ecosystem discoverability, not support responsiveness or adoption. Maturity receives 3 because the release and operational documentation provide useful evidence, while our test command failed and live host behavior remains unverified.

