Install verification: Ecosystem plugin: installs via dsh plugin add but declares no dsh.bundle and mounts no plugin capability — the L1-L5 install ladder does not rate it; the security scan still runs.Install verification details →
What it doesAI
DSH plugins that bridge Claude Code, Codex, and Marvis agents, plus a Ghidra reverse-engineering bridge.
- Provides an agent bridge for Claude Code, Codex, and Marvis
- Includes a Ghidra reverse-engineering bridge
- Distributed as DSH plugins for DeepSeek Harness
Capability tags are AI-inferred or rule-extracted from the README (see badge) — vocabulary-limited, not install-verified. AI = inferred; README = literal keyword from the readme.
Installation
dsh plugin --profile web add github:zp2921060653/dsh-pluginsInstall method: GitHub · runtime test inconclusive (L5 unknown)
Verification & Compatibility○ Ecosystem plugin
Grouped by plugin version: each card lists the dsh versions actually tested against that artifact version and their install verdicts (L5 wins). Untested combinations are omitted — absence means untested, never assumed compatible. L1–L3 are a plugin-level static archive (independent of the dsh version — the same verdict on every dsh row): they sit in each row's level grid next to L4/L5, marked "static", while their evidence is stored once per plugin inside that row's "checks & evidence" instead of being duplicated per version.
External credentials (plugin-level): none detected
package manifest not collected — the declared range is unknown.
| plugin ＼ dsh | dsh 0.2.1-rc.2 current | dsh 0.2.1-alpha.4 | dsh 0.2.1-alpha.2 |
|---|---|---|---|
| 8a771d | · | · | · |
| 707b00 | ○ | ○ | ○ |
8a771dlocal2026-10-14
Ecosystem plugin: installs via dsh plugin add but declares no dsh.bundle and mounts no plugin capability — its level reads L5 · Ecosystem plugin.
707b00github○ 3 ecosystem plugin2026-10-19
https://github.com/zp2921060653/dsh-plugins
○dsh 0.2.1-rc.2current2026-10-19Ecosystem plugin
- • Node.js: v24.14.0
- • DSH: 0.2.1-rc.2
- • Plugin artifact: github · https://github.com/zp2921060653/dsh-plugins (version unstated)
ID: dsh-plugins-151@707b00@dsh0.2.1-rc.2 · profile: l4-sandbox
Issued: 2026-10-19 · expires: 2026-10-26
View checks & evidence
not present in web-profile loader inventory
L1–L3 are plugin-level static conclusions (baseline, identical across dsh versions): evidence is stored once per plugin.
○dsh 0.2.1-alpha.42026-10-18Ecosystem plugin
- • Node.js: v24.14.0
- • DSH: 0.2.1-alpha.4
- • Plugin artifact: github · https://github.com/zp2921060653/dsh-plugins (version unstated)
ID: dsh-plugins-151@707b00@dsh0.2.1-alpha.4 · profile: l3-fullscan
Issued: 2026-10-18 · expires: 2026-10-25
View checks & evidence
not present in web-profile loader inventory
L1–L3 are plugin-level static conclusions (baseline, identical across dsh versions): evidence is stored once per plugin.
○dsh 0.2.1-alpha.22026-10-14Ecosystem plugin
- • Node.js: v24.14.0
- • DSH: 0.2.1-alpha.2
- • Plugin artifact: github · https://github.com/zp2921060653/dsh-plugins (version unstated)
ID: dsh-plugins-151@707b00@dsh0.2.1-alpha.2 · profile: l4-sandbox
Issued: 2026-10-14 · expires: 2026-10-21
View checks & evidence
not present in web-profile loader inventory
L1–L3 are plugin-level static conclusions (baseline, identical across dsh versions): evidence is stored once per plugin.
Author badges
Embed the official badges in your README to showcase security & install-test status:
`[](https://www.dsh.so/artifact/dsh-plugins-151/)`Security Report
Two automated layers, neither a manual review: a static scan (per-plugin rule checks) and a model-driven dynamic security audit that covers only the subset static-flagged at warn or above and never executes plugin code.
Static low, dynamic high — stricter wins → overall high.
—aae715db7af1git2026-10-21aae715db7af1same versionclean · gitVet is an AST static scan (npm artifact or git source).
show static findings (12)
- criticalshellExec · shell
forensics-bridge/lib/adapters/rev.js:5async function exec(t, bin, args, opts) { - criticalshellExec · shell
forensics-bridge/lib/adapters/rev.js:20return exec(t, t.radare2, ['-q', '-c', 'iI', binary], opts) - criticalshellExec · shell
forensics-bridge/lib/adapters/rev.js:27return exec(t, t.radare2, ['-q', '-c', 'aaa; afl', binary], opts) - criticalshellExec · shell
forensics-bridge/lib/adapters/web.js:8async function exec(t, bin, args, opts, envExtra) { - criticalshellExec · shell
forensics-bridge/lib/adapters/web.js:17const r = await exec(t, t.trafilatura, ['-u', url, '--output-format', 'markdown'], opts) - criticalshellExec · shell
forensics-bridge/lib/adapters/web.js:26const r = await exec(t, 'py', args, { ...opts, timeoutMs: opts.timeoutMs || 120000 }) - warningchildProcess · code-exec
agent-bridge/lib/adapters/marvis.js:24child = spawn(exe, args, { windowsHide: true, stdio: ['ignore', 'pipe', 'pipe'] }) - warningnetIp · network
agent-bridge/lib/adapters/marvis.js:82const resp = await fetch('http://127.0.0.1:' + st.port + '/health', { signal: AbortSignal.timeout(5000) }) - warningchildProcess · code-exec
agent-bridge/lib/robust.js:68const r = spawnSync('git', ['config', '--global', '--get', 'http.proxy'], { - warningchildProcess · code-exec
agent-bridge/lib/run.js:8try { spawn('taskkill', ['/pid', String(pid), '/T', '/F'], { windowsHide: true, stdio: 'ignore' }) } catch { /* ignore */ } - warningchildProcess · code-exec
agent-bridge/lib/run.js:32child = spawn(command, cargs, { cwd, windowsHide: true, stdio: ['ignore', 'pipe', 'pipe'], env: env || process.env, ...options }) - warningchildProcess · code-exec
forensics-bridge/lib/adapters/net.js:48const child = spawn(command, cargs, { cwd: opts.cwd, windowsHide: true, stdio: ['ignore', 'pipe', 'pipe'], env: opts.env })
show 3 audit findings
- highWrong cmd.exe escaping plus shell:true lets a quote in an argument break out into arbitrary commands
agent-bridge/lib/run.js:15cmd.exe does not treat backslash-quote as an escape: the embedded \" terminates the quoted region, so a following '&' or '|' is parsed by cmd.exe as a command separator. makeSpawnArgs (line 22) then runs the assembled line with { shell: true }. Reachable callers pass untrusted text: forensics-bridge/lib/adapters/web.js clonePage passes the URL to single-file.cmd/monolith, and agent-bridge/lib/adapters/codex.js line 26 passes the raw prompt as argv to codex(.cmd). A URL/prompt such as x" & calc.exe & " executes arbitrary commands with the harness identity (the same Administrator account that ru
- mediumradare2 -c command string is built by concatenation of the target/count parameters
forensics-bridge/lib/adapters/rev.js:33radare2 executes the -c argument as r2 commands split on ';' and its '!' prefix spawns a shell command (e.g. '!calc.exe'). Any target/count value containing ';' or '!' is therefore executed as an r2/shell command with the plugin's privileges, not merely used as a disassembly address. This is reachable whenever the agent forwards an untrusted string (symbol name or address taken from an analysed, possibly malicious binary) into r2_disasm.
- lowCodex adapter can run the bridged agent with --full-auto, disabling command approvals
agent-bridge/lib/adapters/codex.js:27The prompt is passed verbatim as argv (line 26) and, when the config flag is set, codex executes its own commands without asking for approval. Content harvested by the same bundle from crawled pages or analysed binaries can therefore turn an indirect prompt injection into unattended command execution on the host. The agent-bridge cordis.patch.yml comment additionally documents setting claudePermissionMode: bypassPermissions for the bridged Claude agent. Disabled by default, hence low severity.
Shared run.js assembles cmd.exe command lines with wrong escaping (\") and shell:true, so agent-supplied URLs/prompts can inject shell commands on Windows.
Rule: the dynamic audit can only raise risk, never lower it (raise-only, never an exemption); a wrong raise is withdrawn by human review and the tier falls back to the static verdict. Overall verdict = the stricter of the two layers.
Disclaimer: the static scan and the dynamic security audit are both automated analysis, not a security guarantee. Always review what you install.
Activity
Last commit 2026-09-11 · activity: Stable
• Repo created: 2026-08-31
• Stars: ★ 1 · Forks: 1
• Health: Stable — committed within 90 days
Source
Author README
View on GitHub ↗The author's original README from the repository — fetched live, unedited (the English page loads the default README).