portpilot-assistant

Translate natural language into portpilot commands with safe defaults.

Safety Notice

This listing is imported from skills.sh public index metadata. Review upstream SKILL.md and repository scripts before running.

Copy this and send it to your AI assistant to learn

Install skill "portpilot-assistant" with this command: npx skills add lingengyuan/my-skills/lingengyuan-my-skills-portpilot-assistant

PortPilot Assistant

Translate natural language into portpilot commands with safe defaults.

Workflow

  • Classify intent into one of: scan , who , pick , doctor , free , init , config .

  • Resolve target port/range/path from user text; if absent, apply defaults.

  • For the first portpilot call in a new session, run one permission bootstrap to avoid repeated prompts (details below).

  • Execute read actions directly (scan , who , pick , doctor ).

  • For write actions (free , init --force , config migrate ), require explicit confirmation and show details.

  • Return concise conclusion first, then command output details.

Intent mapping

  • 帮我选个端口 / pick a port:

  • portpilot pick --range 3000-3999 --count 1 --lease-ms 20000 --json

  • 扫描当前端口 / scan current used ports:

  • portpilot scan --protocol both --json

  • 查看 3000 端口 / check port 3000:

  • portpilot who 3000 --json

  • 释放 3000 端口 / free port 3000:

  • First portpilot who 3000 --json , then confirm, then portpilot free 3000 --yes --json

  • 扫描冲突 / run doctor:

  • portpilot doctor --json

  • 初始化配置 / init port config:

  • portpilot init --dry-run --json (default preview)

Safety policy

  • Read actions run automatically.

  • Write actions require confirmation.

  • Confirmation payload must include: port , pid , command , cwd , startTime , action .

  • If dependency failure occurs, run portpilot doctor --preflight --json and return install/fix commands.

Permission bootstrap (one-time)

  • On the first portpilot action in a new session, proactively run one lightweight bootstrap command (portpilot doctor --json ) with escalation.

  • Resolve bundled CLI absolute path first (path-independent):

  • Preferred: $CODEX_HOME/skills/portpilot-assistant/assets/portpilot/bin/portpilot.js

  • Fallback for repo-local usage: <repo>/.claude/skills/portpilot-assistant/assets/portpilot/bin/portpilot.js

  • In that one-time escalation request, set:

  • sandbox_permissions: require_escalated

  • prefix_rule: ["node", "<resolved-portpilot-cli-absolute-path>"]

  • Explain that this is a one-time approval to avoid repeated prompts for future portpilot commands.

  • After approval, run requested commands normally without escalation.

  • If bootstrap was not done (or approval was denied), and a read action fails with permission errors (Operation not permitted / netlink denied / lsof denied), rerun once with escalation.

  • Write actions still require explicit user confirmation before execution.

Execution notes

  • Run bundled CLI from this skill first: assets/portpilot/bin/portpilot.js .

  • Fallback to npx only if bundled CLI is missing: npx -y portpilot-cli@1 ... .

  • Always pass --json for machine-readable parsing in skill scripts.

Source Transparency

This detail page is rendered from real SKILL.md content. Trust labels are metadata-based hints, not a safety guarantee.

Related Skills

Related by shared tags or category signals.

General

wechat-archiver

No summary provided by upstream source.

Repository SourceNeeds Review
General

note-creator

No summary provided by upstream source.

Repository SourceNeeds Review
General

md2wechat

No summary provided by upstream source.

Repository SourceNeeds Review
General

obsidian-markdown

No summary provided by upstream source.

Repository SourceNeeds Review