Plugins

WeKnora knowledge retrieval tools for DeepSeek Harness (dsh): semantic search, document reading and RAG/agent answers over your own knowledge bases.

Local-first memory plugin for DeepSeek Harness: layered L1/L2/L3 memory with automatic capture, manual memos_* retrieval tools, and a built-in memory viewer.

Failure-recovery memory: search and record failure-recovery lessons from real engineering sessions, with BM25 + semantic RAG retrieval and a lessons knowledge base.

DBLP and arXiv literature search, BibTeX retrieval and full-text acquisition with session-owned background jobs; an EduWork-maintained fork of SihanLv/dsh-literature.

Local semantic memory and prior-chat retrieval with source references, editable memories, retention, forget controls, undo, and JSON import/export.

NextTavern is a long-form roleplay plugin for DeepSeek Harness 0.1.7-rc.2 Web, with character-card import and creation, worldlines, worldbooks, layered long-term memory, keyword/semantic/hybrid retrieval, character agents, and card or novel export. SillyTavern character-card import is supported; v0.2.9 fixes distribution and installation.

Verifiable research-report engine for DeepSeek Harness with a content-addressed evidence ledger, versioned sealed reports where every claim carries a verification verdict and the manifest hash seals the directory, and retrieval orchestration that reuses the ctx.web and ctx.jobs seams.

Local Git-backed long-term memory for DeepSeek Harness with hybrid CJK/English retrieval, cited context, transactional sync, previews, and rollback.

DeepSeek Harness integration for OpenContext with durable cross-session memory, automatic prompt capture and recall, session summaries, structured insights, and document retrieval.

Cross-session long-term memory as a local SQLite store: dual scopes (user + per-git-origin project), hybrid FTS5 + local-vector retrieval fused with RRF, an evidence gate that forces an explicit sufficiency judgement, provenance audit chains, SM-2 spaced-repetition review, consolidation distillation and decay forgetting, and a nine-view "Memory Palace" sidebar panel. 22 narrow-parameter tools, no external service, no API key, no Docker.

Puts the Markdown documents, source files, images and video that already exist anywhere in the session workspace into the DSH sidebar, ranked by relevance to the current conversation. The default lexical mode matches your recent messages against title, summary and body locally with IDF weighting (BM25) — no model calls, no network, nothing leaves the machine. The list comes from scanning the workspace rather than from a recent-files list, so restarting DSH or opening a new session does not empty it. Results are organised into a three-tier reading order (primary / supporting / related) with a verifiable reason on every row — which field matched, which document cites this one — and one click opens an in-place preview: Markdown as rendered text, relative-path images actually drawn, source code with the host's own syntax highlighting, video streamed over HTTP Range. The same Context Pack is exposed to the agent as a knit_docs tool. Later versions added read evidence and lifecycle tracking (attribution is frozen at the moment of reading), pin / exclude (exclusion applies before retrieval, and is undoable), relation evidence (references / imports / tests, with line numbers) and an incremental index reused across restarts. Since v0.23 there is also an optional hybrid mode: it is off by default and only runs once you configure an Ollama or OpenAI-compatible embedding provider yourself. When it runs, what leaves the machine is your query text plus each candidate file's title and summary — a Markdown summary is an excerpt of the body's first paragraph (at most 60 characters), a code file's summary is its workspace-relative directory, and an image or video contributes its file name; full bodies and absolute paths are not sent, and a loopback provider keeps everything on this machine. Since v0.24 it also associates uncommitted changes: a read-only git status (fixed argv, no shell, whitelisted child-process environment) combined with the same reference / import / test / mention evidence shows which files a change is likely to drag along, and it is off by default. The benefit of semantic retrieval still awaits a real-provider run, so no recall improvement is claimed.

Reflective self-evolution plugin for local Qwen3.8 27B/3.8 Flash and online APIs: a coding-loop convergence steerer curbs Qwen3.8 27B/Flash repeated-verification deep-thinking to converge faster and cut thinking time; a five-state delivery guard detects and corrects stuck turns and infinite loops; model-agnostic host logic, mostly compatible with online APIs; per-use continual self-improvement with counter-reason self-checks, critic escalation, inverted-index memory retrieval, coding-loop convergence steering, and a five-state delivery guard.

File-based cross-session memory: plain Markdown memory files with automatic retrieval injection, format-compatible with Claude Code's CLAUDE.md and memory files, so migrating is a manual file copy.

A single-character agent that lives across QQ (via SnowLuma / OneBot 11), Bilibili and Douyin (the Douyin channel needs an always-on Windows desktop): persona cards with slowly evolving traits, per-person memory with Ebbinghaus decay and correction, a virtual clock driving a daily routine, proactive messages, auto-extracted memories, and a ten-tab settings panel covering retrieval, persona, routine, memory inspection, channels, habits, models and usage. QQ groups are deny-by-default: a group is answered only when it is on the allowlist or allGroups is explicitly switched on, and it still has to @-mention the bot. Private chats answer QQ friends only. Bilibili/Douyin auto-reply is off by default, limited to the owner, allowlist and mutual follows on Bilibili, and to the owner and allowlist on Douyin unless reply-to-everyone is explicitly enabled. Vendors MIT-licensed code from dsh-qq-bridge and dsh-douyin-bridge, credited in-source with their LICENSE files; it does not need either installed alongside.

Turns local markdown and text documents into a queryable knowledge base with hybrid semantic and keyword search, citation verification, and source injection.

Token optimization for DeepSeek Harness with searchable MCP tool discovery and lazy loading, configurable compaction controls and status reporting, tool-output reduction, long-text vision summarization, and opt-in conversation archiving with paginated retrieval.

Long-term memory for DeepSeek Harness: multi-channel retrieval and an evolving memory store.

Long-term memory for DSH: conversations auto-distilled into atomic facts, scene summaries and a persona profile, with relevant memories injected before every model step. Zero-config out of the box — hybrid BM25 + vector retrieval, optional offline local embeddings, chat/work family separation.

Semantic retrieval, resource management, auto-recall (user + agent dual spaces) and session memory for DeepSeek Harness: memsearch/memfind/memread/membrowse/memgrep/memglob over viking:// URIs, memadd/memremove resource management, memqueue observer status, memcommit session capture, and memlearn skill playbook minting.

Cross-session persona and diary for DeepSeek Harness: each turn can leave a first-person diary entry with a short retrieval index, and the next conversation injects the persona, a hand-note file you write yourself, and a dated batch of past entries retrieved by searching the diary text itself; anything older is found through feiyu_recall using that index.

Automatic context compression for DeepSeek Harness: compresses tool outputs before they reach the model and keeps every lossy compression reversible via CCR retrieval tools.

Same-origin in mechanism with the two loudest lines in context compression. The code-skeleton gate follows the skeletonization approach of Headroom (Apache-2.0), whose published headline is 20% fewer tokens for coding agents and 60–95% fewer tokens for JSON, same answers. The estimator channel follows TokenPilot (arXiv:2606.17016), which reports up to 60% lower cost for long-session agents. Both figures are theirs, quoted as-is; this plugin ships no benchmark of its own and claims no reduction of its own. What it adds for DeepSeek Harness: choose a compression profile, set the Auto Compact trigger level and toggle code-skeleton compression from one settings section, with exact DeepSeek V4 tokenizer measurement, same-revision count verification, and fail-open behaviour that keeps the original tool results on unsupported models. Within a turn it supports skeletonized information compression with a low original-text retrieval rate; at the tail of a turn it supports a turn-tail intent summary implemented purely by content substitution; context content stays safe, and the whole process runs inside the official JSONL compression.

All-in-one local memory: JSONL files per scope (session, project, global, or a custom name) with per-scope dedup, five model tools (memory_save, memory_search, memory_list, memory_get, memory_forget), automatic recall and hard rules injected into the agent pre-step, and a settings panel. Retrieval combines lexical IDF with a local embedding backend, which the panel can install and manage itself.

Experience memory and reflection for DSH agents: automatically extracts candidate lessons after tasks and recalls user-reviewed lessons in relevant later tasks. Supports source provenance, project-scoped retrieval, conflict handling, usage feedback and expiry review; capture and recall have separate controls, and writing lessons to AGENTS.md requires preview and explicit application. A step-by-step first-run guide (4 steps) now opens inside its settings panel.