dsh-plan-lattice
Native continuity recovery for long-running DeepSeek Harness agents
388 results
Native continuity recovery for long-running DeepSeek Harness agents
Context compression tool (context_compact): the agent writes the replacement checkpoint itself and hands it to the host compaction engine, which skips the LLM summarizer call; automatic compaction stays on the official engine.
Structural code MCP server and batch CLI for compact TypeScript/JavaScript reads and hash-bound AST edits.
Zero-dependency MCP client bridge for DeepSeek Harness: connects stdio/HTTP MCP servers and auto-registers their tools for the agent.
Adaptive Context Plane core bundle: Evidence Ledger + Governance + Context Composer + User Model (explicit)
Observe-only provenance ledger over public DeepSeek Harness runtime evidence
Select text in a conversation to get a floating toolbar: copy, quote-reply, ask/explain/translate/summarize, and a /btw side-question console (context-only answers that never enter the conversation). AI actions reuse the current session by default, with a per-action main-thread/side-channel destination switch.
Git-backed, reviewable long-term context for DeepSeek Harness.
Memory protocol enforcement plugin for DeepSeek Harness — forces memory_weave before tool calls, auto-ingests session turns, injects memory context into agent steps. Bridges the opencode-memory MCP server with a hard protocol gate.
In-process graph memory for DeepSeek Harness, backed by TriviumDB.
GPU-aware execution layer for DeepSeek Harness: gpu_status / gpu_exec / gpu_run_bg tools, per-step GPU context injection, automatic CUDA_VISIBLE_DEVICES card selection
Let the model render an image into the DSH web chat flow (QQ/WeChat style) via a show_image tool: the message content keeps only the path text, the image never enters model context, and the browser shows the picture inline, scrollable with history.
DSH context console with trajectory, prompt/skill/MCP/tool inventory, message forging, and conservative session-log recovery.
Ask any line of code its backstory: what it does, and why it's here — grounded in git history and the agent's own session log. Ships a DeepSeek Harness plugin and a standalone MCP server.
Windows minimal agent preset for DeepSeek Harness: a one-line fixed persona, gitbash + str_replace_editor + web_search, no runtime context, no compaction. Installs the preset into the user's agent-presets root.
DSH plugin: 深海摸鱼 —— context 越深潜得越深,答完摸鱼集镭射生物卡 | Deep-sea slacking holo cards
DSH web plugin: adds a one-click Compact button and a New Session button to the conversation context meter panel. Compact submits /compact to summarize older history; New Session starts a fresh session in the same workspace (agent preset and permission settings follow the deployment default).
Cache-stable progressive tool discovery and dispatch for DeepSeek Harness
Codex-style /btw for DeepSeek Harness: ask a side question answered by a conversation-seeded subagent, keeping the answer out of the main context
Inspect, measure, search, and compare the exact provider-neutral context assembled by DeepSeek Harness.
Delegate and monitor DeepSeek Harness work from any MCP agent.
MCP server that searches the Roblox DevForum and official creator docs so AI coding agents can debug Roblox games against real community and engine knowledge.
BrainAgent — brain-inspired plugin for DeepSeek Harness (dsh): a pipeline of heuristic filters, statistical memory stores and context injections; reward-ledger + UCB1 bandit learning loop (RL-lite)
Generic stateless dialogue compression that mimics human memory — forgetting-curve bounded context for LLM conversations, with a ready-to-use adapter module for DeepSeek Harness (DSH).