DSH Session
by DeepSeek
Event-sourced session store for the DeepSeek Harness
CAPABILITY INDEX
Compare memory DSH plugins for cross-session recall, logs, checkpoints, structured persistence, and context recovery.
by DeepSeek
Event-sourced session store for the DeepSeek Harness
by DeepSeek
Semantic session durability checkpoints before model requests and tool side effects
by DeepSeek
Web Session-log export command and shared download dialog
by DeepSeek
JSONL durable session persistence backend for the DeepSeek Harness
by DeepSeek
Session-projection seam: the merge-extensible projection type table, the provider contract, and the ctx.sessionProjections registry serving whole current values of log-derived per-session state
by DeepSeek
Persisted projection cache (ctx.sessionProjectionCache): durable per-session projection checkpoints over the domain data form, throttled write-behind, and the cold-read ladder (cache row + persistence tail replay)
by DeepSeek
Whole-log conversation counts and wall times projection (sessionStats) for the DeepSeek Harness
by DeepSeek
OpenTelemetry backend for the DeepSeek Harness telemetry seam: hands captured session records to the OTel JS SDK's log pipeline
by DeepSeek
Log-backed session title service and provider registry for the DeepSeek Harness
by ohtokaah
Cross-session recall and proactive memory with bounded, progressive context injection.
DSH CATEGORY GUIDE
Memory DSH plugins preserve useful context across sessions through recall stores, logs, projections, checkpoints, or structured persistence. They differ in what is saved and when it is retrieved. A memory plugin should improve continuity without silently injecting excessive or outdated context into every task.
Compare storage location, retention, retrieval rules, context limits, export options, deletion controls, and whether prompts or tool results are persisted. Distinguish active recall from simple session logging, because the privacy and runtime effects are different.
Use test conversations first and inspect the stored files or database. Verify how memory is cleared, what happens after a schema change, and whether sensitive content can be excluded. Keep backups before migrating an existing memory store.
FAQ / 02
Choose active memory when earlier context must be retrieved into new work. Choose logging when the main need is audit, export, statistics, or recovery without automatic context injection.
It varies by plugin: local JSON or JSONL files, SQLite, application storage, or an external service are all possible. Verify the exact path, retention policy, and deletion procedure before using real conversations.