dsh-clear-tool-results
Run the following command in DeepSeek Harness:
dsh plugin install stultuss/dsh-clear-tool-results
Paste the following prompt into your AI chat to install this plugin:
Visit https://github.com/stultuss/dsh-clear-tool-results for details, then run dsh plugin install stultuss/dsh-clear-tool-results to install.
About this plugin
In long DSH conversations, every tool result—bash output, file reads, API responses—keeps sitting in context, inflating token costs and adding historical noise that distracts current-turn reasoning.
dsh-clear-tool-results archives each round's raw tool results into the session's tool-result-logs directory the moment the turn ends, then replaces the in-context copy with a short placeholder so the next turn starts with a leaner prompt. A registered read_tool_result_log tool lets the model retrieve the archived data on demand by turn number or timestamp, with no extra user command required. Archiving is idempotent, so enabling the plugin mid-session or after a restart will automatically backfill any previously missed rounds.
Ideal for developers running tool-heavy multi-turn sessions in DSH who want to shrink context size without losing history. The plugin relies solely on built-in Node modules, supports both legacy and current Harness cores, and coexists with the built-in compaction mechanism.
Screenshots
Use Cases
- Tool-heavy long sessions where context tokens keep growing turn by turn
- Compressing historical tool output while allowing the model to retrieve it on demand
- Backfilling missed archive rounds after re-enabling or restarting the plugin
Best For
- Developers running tool-intensive multi-turn DSH sessions
- Production users who need to keep token costs in check over long conversations
- Advanced users combining built-in compaction with custom memory strategies
Related Plugins
Traceable, searchable cross-session memory for AI agents that turns conversation knowledge into a typed knowledge graph and recalls relevant subgraphs instead of replaying full history, natively integrated with DeepSeek Harness.
Gives DSH AI cross-session long-term memory, to-do and skill management, plus multi-session orchestration, external AI delegation, and an infinite canvas that grows with you.
dsh-mnemon is a three-tier, pluggable, Agent-driven memory system for DeepSeek Harness, combining Runtime memory, Project Documents, and replaceable Memory Spaces with nine long-term providers.
An opinionated, zero-infrastructure file-based memory protocol: human-readable markdown files plus a bounded always-loaded index, curation discipline, and no database or embeddings.