pilot-sandboxBring a Pilot Protocol node online from a network-restricted agent sandbox (Meta Muse and similar hosted VMs): no outbound UDP, poisoned DNS for the Pilot hostnames, and HTTPS CONNECT through an authenticating egress proxy as the only way out. Ships a transparent SNI router plus a mount-namespace hosts trick so pilot-daemon runs in compat mode without touching the TLS handshake. Use this skill when: 1. pilotctl daemon start hangs or the daemon never logs "daemon registered" inside a sandbox, container, or hosted agent VM 2. HTTPS_PROXY is set and direct TCP to registry.pilotprotocol.network fails or resolves to a blackhole address (198.18.x.x) 3. You are setting up Pilot inside Meta Muse's dedicated VM 4. Compat mode alone (-transport=compat) still cannot reach the registry Do NOT use this skill when: - Plain compat mode works (UDP blocked but direct TCP/443 allowed): just run pilot-daemon -transport=compat, see the firewalls doc - The daemon is already registered (pilotctl --json info succeeds) - You cannot get root or CAP_SYS_ADMIN (unshare -m needs it)
Install via ClawdBot CLI:
clawdbot install teoslayer/pilot-sandboxGrade Fair — based on market validation, documentation quality, package completeness, maintenance status, and authenticity signals.
Accesses system directories or attempts privilege escalation
/etc/hostsCalls external URL not in known-safe list
https://pilotprotocol.network/docs/firewallsAudited Sep 30, 2026 · audit v1.0
Generated Sep 29, 2026
An agent team running inside Meta Muse hits the documented wall: outbound UDP is blocked, DNS for pilotprotocol.network resolves to 198.18.x.x blackholes, and only an authenticating HTTPS CONNECT proxy is allowed out. Using this skill, they stand up the SNI router plus a mount-namespace /etc/hosts override so pilot-daemon registers in compat mode without touching the TLS handshake. The recipe is reused after every VM restart because no supervisor exists.
Platform engineers running agents in containers with read-only /etc/resolv.conf and /etc/hosts bind mounts need Pilot for service discovery and relay traffic. The skill routes registry and beacon SNI hostnames through the corporate egress proxy, keeping TLS end-to-end so certificate verification survives. They wrap the recipe into the container entrypoint so registration happens at boot.
An SRE sees `pilotctl daemon start` hang and `daemon registered` never appear in daemon.log. They apply this skill's diagnosis flow: check whether HTTPS_PROXY is set, whether direct TCP to registry.pilotprotocol.network blackholes to 198.18.x.x, and whether plain compat mode is insufficient. The SNI router log plus TLS trust mode (system vs pinned) isolates the failure within minutes instead of six dead ends.
A swarm of agents needs the Pilot trusted directory and beacon relay path for peer ping and trust handshakes, but every non-proxy path is killed by the sandbox network guard. The skill's compat-mode daemon tunnels registry TLS and beacon WSS through the proxy on port 443, enabling `pilotctl ping` through the relay. Verification uses info, trusted list, and ping to confirm both paths work.
Teams that fell back to -registry-trust=pinned because the sandbox lacked a CA bundle must re-fetch the registry leaf fingerprint roughly every 60 days when Let's Encrypt renews. This skill documents the pinning fallback, the PILOT_REGISTRY_FINGERPRINT env var, and the proxy-based fingerprint fetch snippet. The operation keeps a node online while system trust is unavailable.
A hosting or platform vendor sells turnkey Pilot node onboarding for customers whose agent VMs block UDP and force all traffic through an authenticating proxy. The service packages the SNI router and mount-namespace recipe as a boot-time agent, so customers get a registered node without writing any networking code.
Specialist consultants harden agent environments for outbound-restricted networks, diagnosing proxy/CONNECT and SNI routing issues and integrating pilot-sandbox into CI and container entrypoints. Engagements cover TLS trust policy, fingerprint rotation, and mount-namespace permission reviews.
The skill is AGPL-3.0 licensed and freely installable via ClawHub, while the maintainer monetizes premium dashboards, health checks, and automated fingerprint-refresh tooling. Paid tiers monitor daemon registration, proxy tunnel health, and TLS trust expiry across fleets of sandboxed nodes.
💬 Integration Tip
Verify the three constraints first (UDP blocked, DNS poisoned, HTTPS CONNECT-only proxy) and confirm you have root or CAP_SYS_ADMIN for `unshare -m`, since without it the hosts-override step fails. Prefer -registry-trust=system when a CA bundle exists and only fall back to pinned fingerprints with a recurring 60-day rotation reminder.
Scored Sep 30, 2026
Security-first skill vetting for AI agents. Use before installing any skill from ClawdHub, GitHub, or other sources. Checks for red flags, permission scope,...
Security scanner for AI agent skills. 9 built-in detection signatures. Identifies secrets, unsafe execution patterns, and prompt injection. Sub-50ms results.
Wallet anti-theft guard. One-click scan for high-risk wallet approvals to protect user assets. Use when a user asks for a wallet security check, wallet healt...
Comprehensive security audit for an agent's full skill stack. Chains scanner, differ, trust-verifier, and health-monitor into a single assessment with priori...
GEO Audit — AI Search Visibility Checker for ChatGPT, Perplexity, Claude & Gemini. 29-point GEO readiness checklist: robots.txt AI crawler access, Index...
Senior SecOps engineer skill for application security, vulnerability management, compliance verification, and secure development practices. Runs SAST/DAST sc...