dsh-stack
Run the following command in DeepSeek Harness:
dsh plugin install weivwang/dsh-stack
Paste the following prompt into your AI chat to install this plugin:
Run dsh plugin install weivwang/dsh-stack in DeepSeek Harness to install this plugin; the source is available at https://github.com/weivwang/dsh-stack
About this plugin
A plugin list is not a working environment. Bundle order changes composition, version drift changes behavior, and the profile patch is what made the setup useful in the first place. dsh-stack captures that full contract—ordered plugin bundles, exact registry versions and commit-pinned Git sources, a portable profile-level Cordis patch, the source Harness version, and a whole-file SHA-256 digest—into a single small JSON Stackfile. It can live beside a project repo, a release note, a benchmark directory, or a bug report, and anyone can read it in full before allowing it to touch a profile.
The read path and the write path carry deliberately different authority. inspect validates integrity and portability, plan diffs the desired stack against a target profile, and export captures an installed profile into a new file. Only apply mutates a profile, and it does so behind a closed schema check, a rejection of unsafe specifiers and local paths, a printed install/ordering/patch/secret plan, a required --yes flag, an automatic backup, and a verification-and-rollback step. Credential fields are never written verbatim; they become environment-backed placeholders that inspect lists for you, and the real values are supplied only on the receiving machine. After installation, Harness also registers a read-only model tool called stack_inspect so you can pull the summary or the full redacted JSON at any point during a conversation.
If you need to share one agent configuration across a team, lock the environment for benchmarks and bug reproduction, or keep a team handbook in sync with what actually runs on each machine, dsh-stack is the layer that turns "works on my box" into "verified before it runs anywhere else."
Screenshots
Use Cases
- Share one agent configuration across a team with consistent behavior
- Lock plugin versions and bundle order for benchmarks or bug reproduction
- Review plugin bundles and secret references before publishing a Stackfile
Best For
- Engineers who need reproducible agent environments across machines
- Platform teams managing multiple Harness profiles
- Developers sensitive to bundle order and version drift
Related Plugins
A unified suite combining hot runtime injection, task-aware thinking-mode routing, and a graded session protocol with red-team gates to sustain model diligence across long-horizon inference.
ModLens is a vision plugin for DeepSeek Harness that gives text-only models sight by reading images pasted directly into chat, with zero-config setup and multiple vision engines.
On-demand vision for text-only DeepSeek Harness agents: built-in free keyless vision chain and 14 vision tools, routing image turns as tool calls to vision models with pixel fidelity, no Python needed, one-command install.
Give text-only models in DeepSeek Harness eyes, enabling image Q&A, long-screenshot OCR, UI restoration, and GUI visual tasks.
