Plugins

Context lens: visualizes token usage of every request (context composition, KV cache hit rate) and enforces per-turn / per-session budgets with warning and blocking (npm: `dsh-my-context`).

Anatomy panel for the Agent-Body organs: serves a live vitals dashboard at /anatomy inside the harness itself - organs, heartbeat, healing ledger, self-training counters and the real pulse stream - read straight from the runtime files rather than retyped. Host-only, no client bundle and no build step.

Zero-residence context engine: large tool payloads are evicted out of the live context into a queryable store and only deterministic pointers are left behind, so per-turn prompt tokens stop growing with the transcript. Any masked content is reconstructed verbatim from the session log on demand, by tool-call id or session seq.

The organ kernel of Agent-Body: plugins organized as one body. Each command is turned into a nerve impulse that deterministically innervates the organ which should handle it and names the capability to use; a heartbeat with a variable interval (faster on alerts, slower when idle) pumps operator directives, proprioception and homeostasis warnings system-wide; reflex arcs fire with zero model calls; failures go through deterministic attribution (missing tool / arguments / permission / timeout / network), a prescription, and a re-check loop, with argument errors never auto-retried. A missing organ is compensated by the one with the highest capability overlap, and tool schemas are gated by the current intent (84.7% measured reduction, reproducible in-repo). This entry is the kernel plugin package inside the agent-body repository.

The memory organ of Agent-Body: after 2 minutes of silence it enters light sleep and after 5 minutes deep sleep, where deterministic rules distil real experience into memory cards (pitfalls, playbooks, hotspots, unresolved issues, facts). Incoming commands recall related cards by keyword and inject them into context; cards unused for long decay on a half-life and are archived rather than deleted. It also de-noises body-wide homeostasis warnings: repeats inside a window are counted instead of re-printed, and a warning that keeps recurring without ever healing converges into a known steady-state offset and goes quiet. Consolidation and recall make no model calls.

Plugin governance: new/updated plugins stage in a candidate area first, then hot-mount after startup — success promotes, failure auto-disables, repeated failures freeze, one-click safe mode, sidebar diagnostics panel (npm: `dsh-my-guardian`).

Browser notifications and a beep when a session turn ends, the agent asks a question or approval is pending; clicking the notification jumps to the session; remote trigger endpoint and SSE channel (npm: `dsh-my-notify`).

Provides a better model experience, including but not limited to: filling model parameters in multiple ways, improving model compatibility, remembering reasoning efforts, subagents following the parent's model and reasoning effort in real time, and AI self-adjusting reasoning effort.

Task reliability: auto-retry on model timeouts/failures, auto-continue unfinished tasks, completion verification, thinking-repeat interruption, resume after sleep or restart, autopilot mode and a remote trigger endpoint (npm: `dsh-task-reliability`).

Per-call cache billing inside the context-meter popover: current step, round and session totals with hit/miss/output tokens and amounts, automatic Beijing working-hour peak/off-peak pricing, per-model rates, and third-party relays billed too.

A mid-run steer input for DeepSeek Harness Web: while the agent works, an input bar appears above the composer to inject modification requests into the next decision via agent.steer(), without stopping the conversation. Drafts persist per session, unread injected messages can be recalled from the bubble, and an optional guiding prefix is supported.

DSH sidebar file-activity tab: records file reads/creates/modifies by the agent and the sidebar (including file ops hidden inside bash commands), per-session isolation, survives restarts, click a file for a floating preview (npm: `dsh-file-activity`).

A self-contained second brain for DeepSeek Harness — auto-captures durable dialogue knowledge into a local Markdown vault (custom dedup, CJK search, git-friendly), with a graphical knowledge base and an interactive knowledge graph (legend filter, time-dimension coloring, box-select, merge, delete, PNG export/import).

Native Rapid-MLX provider for DeepSeek Harness (dsh) — dsh reads model facts from the server instead of your settings.yaml.

DSH bundle that registers 13 native codegraph_* tools wrapping the codegraph CLI, so agents bootstrap, maintain, and query a pre-indexed code knowledge graph (symbol query, explore, node trail, callers/callees, impact, affected tests) instead of grepping files.

Role-based Codex / Claude Code / ACP subagent providers for the DeepSeek Harness — continuable children, durable session recovery, per-role product permissions, and delegation with a permission ceiling.

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

Animated mood icon: swaps the favicon, sidebar wordmark and hero headline for bloub expression animations that follow session state (working / awaiting input / done / idle), with 8 shapes, 12 colors and custom wordmark texts configurable in Settings.

Presentation skill and tools: one sentence or one document to an HTML slideshow plus editable PPTX, seven layouts (incl. quote and native PPTX tables), speaker notes (HTML S-key panel and native notes slides), five built-in themes, default theme/language config and bilingual support. A step-by-step first-run guide (4 steps) now opens inside its settings panel.

Terminal-style input history for the web composer: edge-first arrow-key recall with exact draft/caret restore, browser-local persisted history, Ctrl+R reverse search, and sliding-context awareness; 0.5.0 adds a smart input layer — cross-session snippets (/save, /load), prompt templates with variables, reuse insights, and compaction-summary highlighting.

Directed fuzzing driven by an LLM agent: from a target repository and a bug (CVE description, patch, crash trace, or file:line), it runs a PLAN, IMPLEMENT, EXECUTE, REFLECT loop in which the model writes Python input generators, until an input triggers the campaign's stderr-pattern oracle or the campaign stops.

LLVM static-analysis plugin that builds whole-program bitcode for C/C++ projects and runs kernel-analyzer (KAMain) reachability analysis on it, providing kanalyzer_* tools to the agent and a ctx.kanalyzer service to other plugins.

Injects one system-prompt section so the agent's chain-of-thought, planning and tool-invocation reasoning stay in Simplified Chinese. Zero external dependencies - clone and link-install from any local directory, no npm build or publish required.