dsh-repeat-stop
Run the following command in DeepSeek Harness:
dsh plugin install 173787247/dsh-repeat-stop
Paste the following prompt into your AI chat to install this plugin:
Run dsh plugin install 173787247/dsh-repeat-stop in your DeepSeek Harness terminal; the full source is at https://github.com/173787247/dsh-repeat-stop, then restart dsh web to activate.
About this plugin
The official repeat-tool reminder only advises; it never blocks. An agent can keep spinning on the same failing call, burning time and tokens with zero progress. dsh-repeat-stop closes that gap: once the same tool with the same arguments has been called consecutively up to a threshold, the next call is hard-denied and the Trajectory view shows dsh-repeat-stop: blocked.
By default six identical calls are allowed and the seventh is rejected. Changing the arguments, switching to a different tool, or a new user message all reset the counter. The threshold, exclusion list, and include list are all configurable, and the plugin pairs naturally with dsh-tool-budget, which caps the session-wide tool total rather than consecutive repeats.
If your DeepSeek Harness workflows involve multi-tool orchestration and the agent keeps getting stuck in a retry loop on the same call, this plugin is the lowest-cost brake pedal available. It introduces no new tools, changes no existing flow, and takes effect the moment you restart dsh web.
Use Cases
- Agent stuck in a retry loop calling the same tool with identical arguments
- Multi-tool orchestration where one tool keeps getting invoked repeatedly
- Pairing with dsh-tool-budget to cap both consecutive repeats and session-wide tool totals
Best For
- Developers running multi-tool workflows on DeepSeek Harness
- Operators or debuggers hitting agent retry-loop issues
- Teams needing fine-grained control over agent tool invocation behavior
Related Plugins
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.
Packages all 87 SKILL.md files from upstream reverse-skill as a DeepSeek Harness plugin that auto-registers them for authorized reverse engineering, penetration testing, and security research.