# Resource Evaluation: Claude Code Remote Control **Date**: 2026-02-25 **Evaluator**: Claude (claude-sonnet-4-6) **Status**: Completed --- ## Metadata | Field | Value | |-------|-------| | **URL** | https://code.claude.com/docs/en/remote-control | | **Type** | Official Anthropic documentation + feature announcement | | **Date** | 2026-02-25 | | **Score** | 4/5 | | **Decision** | Integrate this week | --- ## 1. Content Summary Remote Control allows continuing a local Claude Code session from a phone, tablet, or web browser (claude.ai/code or Claude mobile app on iOS/Android). **Two activation modes:** - `claude remote-control` — new CLI subcommand - `/remote-control` or `/rc` — slash command from an active session **Key architecture:** - Execution is 100% local — the terminal does all the work - Mobile/web interface is a remote window onto the local session - HTTPS outbound only, zero inbound ports, short-lived scoped credentials **Availability:** Research Preview on Pro and Max plans. NOT available on Team, Enterprise, or API keys. **Features:** QR code + session URL for fast connection, auto-enable via `/config`, `/mobile` command for app download links. **Limitations:** 1 remote session at a time, terminal must stay open, ~10min network timeout. --- ## 2. Relevance Score **Score: 4/5 — Very Relevant (significant improvement)** | Score | Meaning | |-------|---------| | 5 | Essential — Major gap in the guide | | **4** | **Very relevant — Significant improvement** | | 3 | Relevant — Useful complement | | 2 | Marginal — Secondary info | | 1 | Out of scope — Not relevant | **Justification:** - Feature entirely new, zero coverage in ultimate-guide.md - Partially tracked in releases tracker (v2.1.51 and v2.1.53) - Introduces a new access paradigm (local execution + remote UI) - HOWEVER: research preview (not GA), limited Pro/Max scope, potentially unstable API --- ## 3. Gap Analysis | Aspect | Doc officielle Anthropic | Notre guide | |--------|--------------------------|-------------| | Remote Control feature | Complete detailed doc | Zero mention in ultimate-guide.md | | `claude remote-control` command | Documented with flags | Missing from cheatsheet | | `/rc` and `/remote-control` slash commands | Documented | Missing | | Security model | Detailed (HTTPS outbound, credentials) | Not covered | | Difference vs "Claude Code on the Web" | Clear comparison table | Not covered | | QR code + session URL | Detailed workflow | Not covered | | Config auto-enable | Via `/config` | Not covered | | Releases tracker | v2.1.51 and v2.1.53 | Already up to date | | `/mobile` command | Mentioned | Missing from guide | --- ## 4. Integration Recommendations ### Files to modify | File | Action | Priority | |------|--------|----------| | `guide/ultimate-guide.md` | New section 9.22 "Remote Control" | High | | `guide/cheatsheet.md` | Add `/remote-control`, `/rc`, `/mobile` to command table | High | | `machine-readable/reference.yaml` | Add `remote-control` entry with target line | High | | `guide/security-hardening.md` | Note on security implications of remote access | Medium | ### Content to document 1. **What is Remote Control** — definition, difference vs "Claude Code on the Web" and vs Teleportation 2. **Setup** — requirements (Pro/Max, /login, workspace trust) 3. **Two workflows** — `claude remote-control` (CLI) vs `/rc` (from session) 4. **Connection** — QR code, URL, mobile app, session list 5. **Config** — auto-enable via `/config` 6. **Security** — outbound-only model, scoped credentials 7. **Limitations** — 1 session, terminal open, 10min timeout, **slash commands don't work in remote UI** 8. **Disclaimer** — Research Preview clearly noted --- ## 5. Challenge (Technical Review) **Score adjusted: 4/5** (down from 5) **Reasons for downgrade:** - Research preview, not GA → risk of documenting content that changes - Restricted Pro/Max scope → majority of pro users (Team/Enterprise) don't have access - Official Anthropic docs already accessible → guide isn't first source for early adopters **Points missed in initial eval:** - Distinction between CLI workflow (`claude remote-control` with `--verbose`, `--sandbox` flags) vs slash command (`/rc` without flags) - Impact for enterprise environments with proxy/VPN/restrictive firewall - `/mobile` command to download Claude app (mentioned in doc, not in our guide) - Comparison with existing alternatives (SSH + mobile terminal, tmux, VS Code remote) **Non-integration risks:** - Short term (1-2 weeks): low, early adopters read Anthropic docs - Medium term (1 month): moderate if feature goes GA - Long term: high if remote-control becomes standard interaction mode **Main risk: over-documenting an unstable preview** → solution: document with clear "Research Preview" disclaimer --- ## 6. Fact-Check | Claim | Verified | Source | |-------|----------|--------| | Available on Pro and Max | ✅ | Official doc + Perplexity (multiple sources) | | Research preview | ✅ | Official doc ("research preview") | | Not Team/Enterprise | ✅ | Official doc ("not available on Team or Enterprise") | | HTTPS outbound only | ✅ | Official doc ("never opens inbound ports") | | Timeout ~10 min | ✅ | Official doc ("more than roughly 10 minutes") | | v2.1.51 first mention | ✅ | claude-code-releases.yaml line 47 | | `/rc` alias for `/remote-control` | ✅ | Official doc | | QR code via spacebar | ✅ | Official doc ("press spacebar to show a QR code") | | Claude app iOS + Android | ✅ | Official doc (App Store + Play Store links) | | "$2.5B annualized revenue" (Perplexity) | ❌ Not verifiable | MLQ.ai article claim, not in official doc. Ignored. | | "29M daily VS Code installations" (Perplexity) | ❌ Not verifiable | Article claim, not in official doc. Ignored. | --- ## 7. Community Articles Found | Source | Title | Date | Type | |--------|-------|------|------| | HelpNetSecurity | "Anthropic's Remote Control feature brings Claude Code to mobile" | 2026-02-25 | Press article | | TechRadar | "Anthropic reveals Remote Control, a mobile version of Claude Code" | 2026-02-25 | Press article | | MLQ.ai | "Anthropic Launches Remote Control Feature for Claude Code" | 2026-02-25 | Tech article | | AwesomeAgents.ai | "Claude Code Gets Remote Control" | 2026-02-24 | Tech article | | YouTube | "RIP OpenClaw!? This New Claude Code Feature is CRAZY!" | 2026-02-25 | Video | | YouTube | "Claude Code Just Destroyed OpenClaw (new /remote-control)" | 2026-02-24 | Video | | Binance Square | "Claude Code Introduces Remote Control Feature for Max Users" | 2026-02-25 | News | **Observations:** Significant community buzz. Several articles compare to OpenClaw (open-source alternative that Remote Control may make obsolete). Significant press coverage for a preview feature. --- ## 8. Community Feedback (FR Slack, 2026-02-25) Source: French-speaking developer Slack discussion (~15 messages, 7 participants) ### Field Insights (Not in Official Docs) | Observation | Impact on guide | |-------------|-----------------| | Bash/AskUserQuestion permissions appear on the phone | Document: mobile serves as an approval terminal | | Slash commands (`/new`, `/compact`) DON'T WORK in remote — treated as plain text | **Major limitation missing from official docs** | | Skills ARE loaded on demand in remote | Positive: skills work | | Works on remote VM (Clever Cloud) in tmux | Advanced use case to document | | Multi-session workaround: multiple tmux tabs, each with own claude session | Pattern to document (bypasses single-session limit) | | Server setup: 6-8 Claude sessions in tmux, no interruption while traveling | Advanced architecture to document | | happy.engineering = main open-source alternative now "obsolete" | Mention in comparison | ### Security Concerns (Senior Devs) | Concern | Analysis | |---------|----------| | "C'est une sacree RCE qu'ils introduisent là" (Frédéric C.) | True: session URL = live access key to local terminal. Real attack surface. | | "Mon CISO va bientôt me challenger" (Mat F.) | Enterprise implication: even on personal plan, if used on corporate machine | | Per-command permissions limit risk but "an attacker would say yes to everything" | Valid nuance: mobile approval guards against unintentional actions, not active attacker | | Suggestion: use on "hardened trusted workstation à la Chromebook" | Best practice to document | ### Overall Sentiment - **Strong enthusiasm**: "KILLER FEATURE", "finally", "here it is" - **Immediate adoption**: several tested it within hours of announcement - **Alternatives made obsolete**: happy.engineering, SSH/VPS hacks, OpenClaw - **Perceived maturity**: "quite limited for now" (Hubert), single-session, no slash commands - **Adoption score**: 8/10 (enthusiasm) but 5/10 (current maturity) --- ## 9. Final Decision - **Final score**: 4/5 - **Action**: Integrate this week with "Research Preview" disclaimer - **Confidence**: High (official doc verified, community feedback confirmed, multiple press sources) ### Added Value vs Official Doc Thanks to community feedback, the guide can provide **original content** not in Anthropic docs: 1. **Slash commands limitation in remote** (`/new`, `/compact` treated as text) — undocumented 2. **tmux multi-session pattern** to bypass single-session limit 3. **Permanent server architecture**: tmux + remote VM + remote-control 4. **Community security analysis**: RCE surface, CISO implications, best practices (hardened workstation) 5. **Alternatives comparison**: happy.engineering, OpenClaw, SSH tunnels, VPS setups