dsh-approval-llm
Run the following command in DeepSeek Harness:
dsh plugin install Letter2025/dsh-approval-llm
Paste the following prompt into your AI chat to install this plugin:
In DeepSeek Harness, install the plugin using the command 'dsh plugin install Letter2025/dsh-approval-llm' from https://github.com/Letter2025/dsh-approval-llm.
About this plugin
In DeepSeek Harness, permission approvals typically require manual human intervention, which can become a bottleneck in automated workflows. The dsh-approval-llm plugin addresses this by introducing a model approval mode that uses a separate reviewer model to automatically handle approval requests, enhancing efficiency. Its core capabilities include: the reviewer model can decide to ALLOW, DENY, or ESCALATE requests, activating only under configured permission presets; it supports fine-grained tool list controls such as allowlist, denyList, and humanOnlyList; and it features a circuit breaker mechanism to prevent excessive denials from the model. This plugin is ideal for developers and administrators looking to automate low-risk workflows while reducing manual oversight, with important security considerations noting that AI reviewers can be vulnerable to prompt injection, making it suitable for lower-risk scenarios.
Use Cases
- Automate approval processes for low-risk tool invocations.
- Quickly validate tool permissions in development and testing environments.
- Reduce manual approval burden and enhance workflow efficiency.
Best For
- Developers using DeepSeek Harness.
- Operations personnel responsible for system security management.
- Teams needing automated approval policies.
Related Plugins
A network security red-team benchmark plugin for DeepSeek-V4.1/Flash featuring dual-layer kernel injection and zero-tool architecture for controlled model safety assessment.
A security red-team prompt plugin for DeepSeek-V4 that uses code-formal mapping and zero-trigger system prompts to enable unconditional output, intended for authorized testing and research only.
Adds an Auto approval mode on the official workspace-write sandbox, classifying semantic risks, asking when ambiguous, and denying destructive operations.
Undo/rollback system for DSH: config & plugin snapshots, one-click undo/redo/restore, message-level revert, secret masking, safe mode, and offline WebUI/GUI/CLI rescue tools for when DSH won't start.