dsh-continual-harness
Continual learning and self-improvement harness plugin providing persistent memory, automated review, and rollback.
Install
$ dsh plugin --profile <name> add dsh-continual-harnessPlugin 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
- Injects state and memory context incrementally into agent pre-step workflows based on digest changes
- Executes automated LLM review, planning, and refinement on turn intervals or session compaction
- Provides harness_refine tool and /refine slash command with atomic apply and rollback capabilities
- Materializes applied skill contracts into real DSH SKILL.md bundles
- Executes code-owned A/B benchmarks using frozen test cases and baseline reference snapshots
- Supports session wrap-up recommendations and injection telemetry logging
Useful For
- Persisting learnings, error patterns, and instructions across multiple agent sessions
- Automatically refining agent prompt directives and skills during long-running tasks
- Running controlled A/B benchmarks before adopting prompt or skill modifications
- Safely rolling back problematic agent memory or contract updates
Who It Fits
- Developers building autonomous or self-evolving agents in DeepSeek Harness
- Engineers requiring cross-session knowledge persistence and structured agent memory
- DSH power users optimizing agent prompt behaviors with regression testing
Documented Limitations
- Real end-to-end evaluation requires configured LLM capability and API keys
- Concurrent refinement writes follow a last-writer-wins model without multi-process serialization
- Compaction triggers silently skip when the session compaction capability is absent
- Content-shrink guard is not yet implemented; only per-commit entry growth is capped
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
- jasen215/dsh-continual-harness
- Registry source
- GitHub · dsh-plugin topic
- Registry classification
- Plugin
- Source checked
- d4e336a · 2026-09-14
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
- 9GitHub stars
- Forks
- 1
- Open issues
- 0
- Last commit
- 2026-09-10
- Last release
- 2026-09-10
Related DSH Plugins
Ranked by overlapping capabilities, use cases, plugin type, categories, and DSH profile.
Multi-agent team collaboration plugin for DeepSeek Harness with DAG task scheduling, mailbox messaging, and an interactive Web UI monitor.
View plugin →Persistent multi-model Agent teams and bounded DAG workflows with GUI controls for DeepSeek Harness.
View plugin →Creative fiction and short-drama production workbench plugin for DeepSeek Harness.
View plugin →Second-model AI auto-review for DeepSeek Harness approval requests with fail-closed safety and session audit.
View plugin →