openma-ai/backchat

C2

Local-first desktop workspace for ACP agents—DeepSeek Harness, Claude Code, Codex, Gemini CLI, OpenCode, Hermes, and OpenClaw—with projects, tools, files, browser, and sessions.

★ 5+ · openma-ai/backchat source on GitHub · this plugin in the registry

openma-ai/backchat is a DeepSeek Harness plugin rated C2 — one powerful capability or sensitive behavior. It executes system commands, decodes base64 payloads, starts a network server, reads credential-class env vars.

What it can do

CapabilityFlagEvidence
executes system commandsexec×38 src, e.g. scripts/prepare-bundled-npm-runtime.mjs:3, scripts/prepare-bundled-npm-runtime.mjs:3
decodes base64 payloadsbase64_decode×2 src, e.g. src/main/browser-plugin-extension-adapter.ts:477, src/main/ui-fs-broker.ts:278
starts a network servernet_server×12 src, e.g. src/main/browser-extension-http-bridge.ts:98, src/main/browser-harness-mcp.ts:49
reads credential-class env varstoken_envBACKCHAT_DEEPSEEK_TEST_KEY

Outbound domains

agentclientprotocol.com cdn.agentclientprotocol.com unpkg.com cdnjs.cloudflare.com esm.sh cdn.jsdelivr.net fonts.gstatic.com fonts.bunny.net opencode.ai fonts.googleapis.com

Environment variables it reads

CLAUDE_AGENT_ACP_BIN CLAUDE_AGENT_ACP_CWD CODEX_PATH npm_execpath CODEX_ACP_BIN CODEX_ACP_CWD INITIAL_AGENT_MODE BACKCHAT_DEEPSEEK_TEST_KEY ACP_AGENT ACP_PROMPT

How to read this

Levels measure capability surface and transparency, not maliciousness. A C3 plugin can be entirely legitimate — a desktop shell genuinely needs subprocesses. The point is that you can see this before installing. See the levels explained and how dsh plugins work.

Findings come from static analysis of shipped code; nothing is executed. Think a flag is wrong? Open an issue — every flag cites the file and line it came from.