dsh-plugin-turn-memory
DeepSeek Harness plugin for turn-level transcript compression, session compaction, and recoverable context-memory rewrites.
Install
$ dsh plugin --profile web add github:vilicvane/dsh-plugin-turn-memoryPlugin Overview & Capabilities
AI-assisted organization based on the public repository snapshot. The content must be grounded in source evidence and does not replace compatibility or security verification.
Key Capabilities
- Compresses completed root-session turns through forked main-model workers.
- Recovers unlanded completed-turn compression after an agent cold resume.
- Provides optional ctx.compaction whole-session compaction with sequential workers and durable checkpoints.
- Offers long-turn handoff reminders and continuation into a fresh turn after compression lands.
- Stores compressed image references as durable memory-image tags and exposes read_memory_image for on-demand reloads.
- Supports configurable token-pressure thresholds, retention, worker budgets, timeouts, and surface snapshot output.
Useful For
- Reduce retained context from completed multi-step agent turns.
- Continue long-running tasks in a fresh turn while preserving completed-work context.
- Replace the default session compaction provider with a configurable turn-memory engine.
- Keep image content available by reference rather than including pixels in every later model request.
- Recover pending completed-turn compression after a Web restart or agent cold resume.
Who It Fits
- DeepSeek Harness users running the web profile.
- Plugin developers and operators managing Cordis profile configuration.
- Users with long or tool-heavy agent sessions that need context management.
Documented Limitations
- The plugin is disabled by default and must be registered in cordis.patch.yml before use.
- Session compaction is opt-in and cannot coexist with another ctx.compaction provider.
- Session compaction setup requires disabling compaction-basic and configuring the profile accordingly.
- Image reload support requires an attachment store and an active model that declares image input.
- End-to-end workflows use a dedicated profile and a configured real model.
DSH Compatibility
Version-specific runtime evidence collected by DSH Plugin. A missing result means we have not tested that combination yet.
Security Signals
Objective signals discovered from package metadata and source inspection. These are not a guarantee that a plugin is safe.
Source & Registry Notes
Public provenance, Registry classification, and the latest source check for this entry, kept separate from runtime verification.
- Source repository
- vilicvane/dsh-plugin-turn-memory
- Registry source
- GitHub · dsh-plugin topic
- Registry classification
- Plugin
- Source checked
- 47db290 · 2026-09-16
This project is independently indexed from public source information. DSH Plugin is not affiliated with DeepSeek or the plugin author. Always check the author repository before installation.
Repository Activity
- GitHub stars
- 1GitHub stars
- Forks
- 0
- Open issues
- 0
- Last commit
- 2026-09-12
- Last release
- No release detected
Related DSH Plugins
Ranked by overlapping capabilities, use cases, plugin type, categories, and DSH profile.
Persistent multi-model Agent teams and bounded DAG workflows with GUI controls for DeepSeek Harness.
View plugin →Composable three-tier memory control plane for DeepSeek Harness with runtime context, searchable documents, and pluggable long-term memory.
View plugin →A standalone, local persistent-memory system for AI agents, with DeepSeek Harness support provided through the separate dsh-mnemon plugin.
View plugin →Local-first AI agent runtime and DSH plugin bundle providing sandboxed sessions, MCP tools, audit logs, and session replay.
View plugin →