dsh-sandbox-escalation-fix
Session-aware sandbox escalation compatibility plugin for DeepSeek Harness
42 results
Session-aware sandbox escalation compatibility plugin for DeepSeek Harness
Rule-gated automatic approval for DeepSeek Harness sandbox escalations with an optional LLM reviewer and native human fallback.
Normalize redundant sandbox requests and malformed justifications in DeepSeek Harness tools
为 DeepSeek Harness 增加介于 Workspace Write 与 Full access 之间的自动批准权限档:例行沙箱升级由分类模型一次性放行,危险或不确定的操作仍转人工审批。An auto-approval permission preset for DeepSeek Harness between workspace-write and full access: routine sandbox escalations are granted once by a classifier, while dangerous or uncertain requests still go to a human.
Security scanner for DSH plugins: static and semantic passes over plugin source, DSH-specific attack-surface rules, npm audit, batch scanning, and an HTML report with per-finding severity and evidence.
CC-style auto mode for DeepSeek Harness: deterministic deny/allow rules + pre-execute gate + model-agnostic two-stage classifier. Two-state (allow/reject) classifier since 0.8.0. TypeScript rewrite merging dsh-auto-mode v0.4.1 with Nuo-cl/dsh-auto-mode native integration.
AI approval answerer using the unified ctx.llm route with fail-closed local policy checks
DSH compatibility plugin that treats redundant non-escalating sandbox permission fields as a no-op
QuickJS/WASM-isolated WorkflowEngine for running model-written DeepSeek Harness orchestration with bounded resource controls.
Local prompt-injection and secret-exfiltration firewall for DeepSeek Harness
Model-facing typed tools wrapping the frozen, reviewed underseal adapter for the DeepSeek Harness
Agent-decided approvals for DeepSeek Harness: a workspace-write base permission mode where an independent approval subagent judges every sandbox escalation (risky operations are rejected), with a configurable approval model and a per-session audit trail in the conversation window's 审批 tab.
DeepSeek Harness plugin (temporary, pre-upstream-fix): automatically injects a CPython tempfile shim (PYTHONPATH -> sitecustomize) into every confined shell command on the Windows sandbox, so python/pytest tempfile use works with zero extra tools, zero model-context overhead, and zero escalation.
Retrieved ≠ injected: CBDC-gated memory for DeepSeek Harness — decides how memory is USED (use/verify/ignore decisions, feedback learning, full audit); local SQLite + FTS5, bounded, no extra model call
Register models, assist with portraits, and select the Agent model from a secret-free catalog for DeepSeek Harness.
OMV Audit Desk: an evidence-first vulnerability audit workbench for DeepSeek Harness
Seatbelt (libsandbox) sandbox provider for the DeepSeek Harness sandbox seam on macOS: a cordis plugin that replaces @deepseek-ai/dsh-sandbox-local's seatbelt rung via a prebuilt seatbelt-run launcher — no dsh source changes, no PR required
NeevSandbox execution providers for DeepSeek Harness
Mingleng mcpguard for DeepSeek Harness 鈥?the first security plugin for DSH. Scans skills and MCP configs for prompt injection, homoglyphs, hidden Unicode, dangerous shell and credential leaks.
Fail-closed DSH compatibility guard for redundant GPT/Codex sandbox escalation arguments
Strip sandbox_permissions / justification from model tool-call arguments so sandbox escalation is never triggered when the session already has sufficient permission.
Project DeepSeek Harness tool, approval, and result facts into a bounded action lifecycle ledger.
Dependency-installation guard for DeepSeek Harness: classifies package-manager commands as ALLOW, ASK, or BLOCK before execution.
Approval timeout for DeepSeek Harness: when a sandbox/escalation approval is not answered within a configurable period, the request is automatically rejected instead of waiting forever.