dsh-session-handoff
Run the following command in DeepSeek Harness:
dsh plugin install realpkuasule/dsh-session-handoff
Paste the following prompt into your AI chat to install this plugin:
Run dsh plugin install realpkuasule/dsh-session-handoff in DeepSeek Harness to install; the source code is available at https://github.com/realpkuasule/dsh-session-handoff . Restart the web profile after installation to activate.
About this plugin
Long-running sessions are everyday reality in DeepSeek Harness: after dozens of turns, context bloats, prefix-cache efficiency degrades, and rework piles up. Switching to a fresh session manually is easy to forget and hard to keep consistent. dsh-session-handoff turns this into an observable, auditable pipeline: the plugin only watches, judges, and triggers, while all actual handoff execution is delegated to the harness-automation CLI. It never duplicates handoff logic.
The core flow has three stages. A projection unit folds live session metrics including turns, steps, LLM latency, title drift, retries, and splice count. When thresholds are crossed and the signal fingerprint has not been seen before, the entry moves to a pending queue. On the next open turn, a deterministic, append-only suggestion is injected, optionally gated by an approval request, and only after approval does the CLI execute the handoff. There are no background timers, no unilateral session termination, no policy-file mutation, and fingerprint deduplication guarantees restart safety.
This plugin suits developers who already run long sessions on the DeepSeek Harness web profile. After installation and a web-server restart, the system automatically senses when a session has reached a natural handoff point, with built-in cacheObs before/after evidence to confirm that prefix-cache integrity is preserved throughout the entire handoff process.
Use Cases
- A multi-hour session bloats and the plugin injects a handoff suggestion on the next open turn
- Title drift hits the threshold in a multi-phase task, triggering a phase-drift hint to prevent context confusion
- Post-handoff cacheObs before/after comparison confirms prefix-cache integrity is preserved
Best For
- DeepSeek Harness developers who run long sessions daily
- Team environments that require auditable and reviewable handoff workflows
- Practitioners focused on prefix-cache efficiency and context lifecycle management
Related Plugins
Memory layer for coding agents that indexes local session history and auto-recalls relevant context before edits, commands, and failures, with no manual search needed.
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.