Second-model AI auto-review for DeepSeek Harness approval requests: a read-only reviewer subagent returns structured allow/deny verdicts with reasons, fail-closed by default, fully auditable from the session log (approval/asked -> autoReview/verdict -> approval/decided).
A dsh plugin that periodically scans reddapi.dev for new Reddit leads matching your one-sentence ICP, dedupes what you've already seen, and writes a dated markdown report. Also forwards 6 read-only Reddit search/lookup MCP tools.
Read-only security & compliance plugin for DeepSeek Harness: prompt-injection detection, Chinese-PII redaction, and local configuration audit with redacted, reproducible reports.
DeepSeek Harness (DSH) plugin: a read-only ledger for the plugins you already have installed — a capability inventory with file:line evidence, declared-vs-detected reconciliation, cross-profile version drift, and a diff of what changed since the last scan.
Read-only performance diagnostics for DeepSeek Harness: session load/restore timing, spill-hit counts, compaction count and trigger, context-injection volume (AGENTS.md/skills/tool-schema token share), and LLM cache hit rate — surfaced via /fast, persisted as reconstructable session events with async sampling off the model path.
File-diff visualization between checkpoint time nodes for DeepSeek Harness: read-only timeline + per-file line diff over dsh-checkpoint-rewind checkpoints, as a /diff command, JSON HTTP API, and GUI panel.
洞察 — read-only insight panel for a DeepSeek Harness profile: where every plugin, service, tool and model came from, which config layer inserted or disabled it, and what is running right now.
DSH web plugin: a Skill Manager page in the settings panel browsing system / user / workspace / preset skills, with file-tree editing, zip import/export, and delete (system skills read-only).
Read-only GitHub access for agents, wrapping several tools. Inside the tool it automatically probes and selects whichever GitHub access method works locally, and tries network proxies on its own. That cuts down the network and parsing problems an agent may hit when accessing GitHub, and the overhead of it repeatedly trying different approaches.
