context-management

Model-driven context management (Active Context Pruning / ACP) for the DeepSeek Harness — the model decides when and what to compress. Ported from billion-context-pi (ranxianglei); acp-kernel reused verbatim. CompactionEngine backend with compress/decompress/search_context/acp_status tools.

Cross-agent plugin for manually excluding and editing AI conversation context, with search, filtering, hide/restore, undo, and original Session preservation.

Branch from any answer and keep branching at any depth. Recursively isolated follow-ups that leave the main DSH task untouched.

为 DeepSeek Harness 打造：选中 AI 回答或思考内容，随时追问、学习、检查与纠偏。

Execution-time drift firewall for long-running DeepSeek Harness agents. Real-Harness tests: unsafe stale mutations 12/12 native -> 0/12; valid controls 7/7 both; post-SIGKILL unsafe continuation 2/2 -> 0/2.

The navigation of AI Agent.

Read-only side chat plugin for DeepSeek Harness (DSH) — ask temporary questions with inherited context while the parent agent keeps running.

Model-driven context-window rollover bundle for the DeepSeek Harness: fresh working context, durable notes, verbatim recent tail, no summarization

Codex-style windowed context management with live notes and cold history search for DeepSeek Harness

DeepSeek Harness 会话健康插件：真实数据的「继续 vs 新开」上下文罗盘——头部徽章 / /compass / context_compass 工具 / 多会话一览面板

Agent instruction manager for DSH — edit and template the AGENTS.md files the harness actually reads (global + per-workspace)

DeepSeek Harness 动态上下文管理插件（Dynamic Context Pruning for dsh），对标 opencode-dcp

该仓库暂未提供项目说明。

Event-based long-term memory for LLM agents — associative surfacing over query RAG. Closed-loop episodic events, budgeted injection, tiered forgetting. Claude Code hooks + DeepSeek Harness plugin, one shared .memory/ format.

该仓库暂未提供项目说明。

Multi-agent frugality defense plugin for DeepSeek Harness: read-ledger dedup, compaction-immune rules, completion gate, cheap-review lane.

Idle-triggered context compaction plugin for DeepSeek Harness: compact a session once it goes quiet past a configurable token floor.

Reusable trajectory-tree context, exact-turn forks, and bounded cross-session recall for DeepSeek Harness

dsh 插件：CTM 上下文/终端管理，把模型上下文变成可见、可编辑、可评分的一等对象

High-fidelity, faithful, bilingual, recursive compaction backend for DeepSeek Harness.

Token-efficient tool-result pruning for DeepSeek Harness: compact previews, targeted recall, and long-session context control.

DSH plugin to A/B test post-compaction We/Let's reasoning with Minimal-first epochs and full tools after the first tool call

Forgetting engine for AI agents — memory TTL, decay, eviction, audit. The opposite of memory programming. DSH plugin + local skill + CLI.

Persistent, searchable, per-project memory for the DeepSeek Harness: decisions, rules, and session context in a queryable DuckDB file, with the rule set injected into every model request — plus a full management UI in the Web Client.

Sleev integration for DeepSeek Harness with route-aware LLM telemetry and context-optimization observability