marq
Health Warn
- License — License: AGPL-3.0
- Description — Repository has a description
- Active repo — Last push 0 days ago
- Low visibility — Only 5 GitHub stars
Code Fail
- rm -rf — Recursive force deletion command in .github/workflows/image.yml
- rm -rf — Recursive force deletion command in scripts/demo-skills.sh
- rm -rf — Recursive force deletion command in scripts/e2e-image.sh
- rm -rf — Recursive force deletion command in scripts/e2e.sh
Permissions Pass
- Permissions — No dangerous permissions requested
No AI report is available for this listing yet.
Universal cyber assistant: ~80 audited Kali tools + an expert skills library, driven by any model over MCP or direct run
Current security expertise for your AI agent
An open knowledge pack of security, GRC and OT/ICS playbooks that is re-checked
against primary sources every 6–8 weeks, plus an optional passive-OSINT tool
server. It works with whatever model and agent you use.
"Is this CVE in our firmware 'actively exploited' under the CRA?" → CRA triage
"Map these pentest findings to IEC 62443, CRA and NIS2." → crosswalk
"Our breach-notification clock under NIS2, GDPR and US law?" → regulation
"Threat-model this payment API." → architecture
"What does this malware sample actually do?" → malware
"Footprint acme.com from public sources." → OSINT
The knowledge pack. About 83 playbooks for practitioners at EU and US
manufacturers. They run from a finding to the obligation it touches (IEC 62443,
EU CRA, NIS2, ISO 27001, NIST, US sector rules) to the board report. Every skill
records the date it was last checked, and Apache-2.0 means it installs into any
agent, including internal ones.
The tool server (optional). About 30 wrapped passive OSINT tools (domain,
infrastructure, people, leaked secrets in public code). They query public sources
and never probe, scan or exploit a target, and every call lands in a fail-closed,
operator-attributed audit log you can hand to a client or regulator.
Start
Pick your path — start simple, add a layer only when you need it:
| I want… | Path | Setup |
|---|---|---|
| cyber expertise inside my agent | skills pack | one install command |
| …the same, over MCP for any client | skills server | one go install, no Docker |
| …plus the ~30 OSINT tools to run | full MCP server | build/pull the image |
| …driven by a local model | full + Pi | image + Pi + llama.cpp |
The skills pack — install into your agent (no server, no Docker)
The ~83 playbooks are a standard Agent Skills pack underskills/, so agents that read SKILL.md load them natively:
# Claude Code (v2.1.292+; on older versions: claude plugin marketplace add
# rhaist/marq && claude plugin install marq@marq)
claude plugin install marq --marketplace rhaist/marq
# Codex, Cursor, Copilot, Gemini CLI, Pi and ~40 other agents
npx skills add rhaist/marq
Each skill carries metadata.verified, the date of its last review sweep. The
pack is re-verified against primary sources every 6–8 weeks; marq skills --stale
lists anything overdue.
Just the skills over MCP — for any MCP client (no Docker)
The lowest-friction way in: add the skills library + advisory tools to any MCP
client and skip the OSINT tool suite entirely. MARQ_SKILLS_ONLY=1 loads only
the in-process knowledge tools (skills, findings, files) — no OSINT
binaries, so it's a single static Go binary you run natively:
go install github.com/rhaist/marq/cmd/marq@latest # ~10 MB, skills embedded
marq skills # browse all 83, by domain
marq skills nis2-dora # read one, no client needed
claude mcp add marq-skills \
-e MARQ_SKILLS_ONLY=1 -e MARQ_WORK_DIR=$HOME/.marq \
-e MARQ_AUDIT_LOG=$HOME/.marq/audit.jsonl \
-- marq serve
Now ask "Map our audit findings to ISO 27001 and SOC 2" or "What's our
breach-notification clock under GDPR and NIS2?" and the model loads the right
playbook (current 2026 standards) and drafts the deliverable. Scripted
end-to-end demo: scripts/demo-skills.sh.
Prefer containers, or already pulled the image? The same mode runs there too — use
the marq-skills entry in mcp.json.example. Add the full tool
suite below when you need to run something, not just reason about it.
The full tool suite
To run the OSINT tools (domain, people, leaked-secret lookups), not just reason, get the image:
# prebuilt — or build from source: docker build -t marq .
docker pull ghcr.io/rhaist/marq && docker tag ghcr.io/rhaist/marq marq
# first check: which build, and how many tools/skills did it come with
docker run --rm marq version
# the catalog — add a name (`marq tools subfinder`) for one tool's schema
docker run --rm marq tools
No Docker Desktop needed — OrbStack (macOS) or Podman / Colima (Linux) run
it just as well (docs/SETUP.md). Then drive it fully local (below)
or point a frontier MCP client at it.
Run it fully local
Your own model, on your box: no cloud, nothing leaves the
host — the recommended setup when investigating people or a client's footprint. Drive marq from
Pi over llama.cpp:
# 1. the model — full flags matter: --reasoning-format keeps chain-of-thought
# out of the reply, --jinja makes tool calls parse (see docs/SETUP.md §2a)
llama-server -hf google/gemma-4-12B-it-qat-q4_0-gguf \
--host 127.0.0.1 --port 8080 -ngl 99 --ctx-size 65536 --jinja \
--reasoning-format deepseek \
-fa on -ctk q8_0 -ctv q8_0 \
--temp 1.0 --top-p 0.95 --top-k 64
# 2. marq in a long-lived container bound to your workspace. The shim ships
# inside the image, so it always matches the marq it drives — no clone.
mkdir -p ~/.local/bin && docker run --rm marq shim > ~/.local/bin/marq
chmod +x ~/.local/bin/marq
marq up ~/work
# 3. one-time Pi config — marq's system prompt. Without it a small model
# narrates instead of driving marq. → docs/SETUP.md §4
mkdir -p ~/.pi/agent && marq prompt system > ~/.pi/agent/SYSTEM.md
# then run `pi`
Then just talk to it: "Footprint example.com: subdomains, emails, exposed hosts," or
"Load the nis2-dora skill — what's our breach clock?" Full local walkthrough
(Pi provider, skill, tuning) → docs/SETUP.md.
Or fall back to a frontier model
For the heaviest reasoning (architecture, GRC, IR write-ups), point a frontier MCP
client at marq instead:
claude mcp add marq -- docker run --rm -i \
--security-opt no-new-privileges:true --cap-drop ALL \
-v marq-audit:/var/log/marq -v marq-work:/work -e MARQ_OPERATOR="$USER" marq
Any MCP client works (Codex, Claude Desktop) → docs/CLIENTS.md.
(No keys or config required; optional
API keys in docs/USAGE.md.)
Passive and purpose-bound. marq queries public sources only and audit-logs
every call; people OSINT is personal-data processing, so record a purpose withset_engagement—docs/SECURITY.md.
What you get
- Execution — ~30 wrapped passive-OSINT tools (domain/infra footprint, people,
leaked secrets in public code), each audit-logged and sandboxed; findings render
tofindings.md/.csv(with optional CVSS + CWE). - Knowledge — ~83
load_skillplaybooks across 13 domains, web-researched to
current (2026) standards, telling the model how to use the tools and what to
do where there's no tool: malware, threat-intel, sec-ops, architecture,
GRC, standards & EU/US regulation, CISO, purple-team, resilience, human factors,
DevSecOps/privacy. Run these standalone withMARQ_SKILLS_ONLY=1— no tool
suite, no OSINT binaries. - Auditable by design — every external-tool call funnels through one choke
point: an append-only,fsync'd audit log, a timeout, and an output cap. File
access is sandboxed to/work+/tmp, and the image runs non-root (the
recommended run args also drop all Linux capabilities +no-new-privileges).
| Category | Tools |
|---|---|
| Domain / infrastructure | theharvester · spiderfoot · subfinder · dnsx · dns_lookup · whois_lookup · asnmap · cdncheck · gau_urls |
| Exposure lookups | shodan_host · shodan_search · censys_view |
| Leaked secrets / metadata | trufflehog · exif_metadata |
| OSINT — people | sherlock · maigret_username · email_accounts · h8mail_breach · phoneinfoga |
Files (sandboxed /work, /tmp) |
list_dir · read_file · write_file |
| Findings & jobs | report_finding · render_report · list_jobs · job_status |
| Session & knowledge | server_info · set_engagement · load_skill |
MCP client ──stdio JSON-RPC──▶ marq serve ┐
terminal agent (Pi) ──bash──▶ pi/marq shim ─┤ both hit the same registry
(docker exec) │
▼
registry.All() ── one Tool list
├─ osint theHarvester, spiderfoot, subfinder, dnsx, shodan, trufflehog, gau…
├─ people sherlock, maigret, user-scanner, h8mail, phoneinfoga
├─ files read/write/list (sandboxed /work)
└─ extras report_finding, render_report, list_jobs, load_skill
│
runner.Run ──▶ audit.jsonl (every call)
load_skill ──▶ skills library (markdown, ~83 playbooks / 13 domains)
Every tool funnels through internal/runner/runner.go::Run — the single
point for audit logging, timeouts, and output truncation.
Documentation
| Doc | What's in it |
|---|---|
docs/SETUP.md |
Per-OS install — skills-only (no Docker), full suite, local model |
docs/CLIENTS.md |
Pick a client/model — Claude Code, Codex, Pi + llama.cpp |
docs/USAGE.md |
Every tool, env vars, API keys, reading the audit log |
docs/SECURITY.md |
Legal/ethical baseline, the guardrail model, hardening |
License
GNU AGPLv3 — see LICENSE. © 2026 marq contributors. Free software;
run a modified version as a network service and you must offer users its source.
The skills pack under skills/ is Apache-2.0 (skills/LICENSE),
so you can install and adapt the playbooks in any agent, including internal ones.
For authorized security testing and education, no warranty — use responsibly.
Reviews (0)
Sign in to leave a review.
Leave a reviewNo results found