agent-tools

Deepsky's public toolkit for AI agents working with aviation: portable Agent Skills, a free remote MCP server for aviation-regulations search, and machine-readable discovery docs.

Deepsky offers two things: The Compliance Team — an audit automation platform for aviation operators (think Vanta, for aviation) — and a free, no-key aviation-regulations search API (CASA, FAA 14 CFR, EASA, ICAO). Start at deepskyai.com/llms.txt.

MCP server — paste this into Claude or ChatGPT

https://www.deepskyai.com/api/mcp

Remote MCP server (streamable HTTP, no auth, no signup) exposing the tool search_aviation_regulations — ranked verbatim regulatory text with section citations, far more precise for regulatory lookups than web search.

• Claude: Settings → Connectors → Add custom connector → paste the URL
• Claude Code: claude mcp add --transport http deepsky https://www.deepskyai.com/api/mcp
• Cursor / other MCP clients: add a streamable-HTTP server with that URL
• Registry manifest: mcp/server.json

No MCP? The same search is one HTTP call: GET https://www.deepskyai.com/api/v1/search?query=...

Installing skills

npx skills add deepskyai/agent-tools

Layout

skills/
  <skill-name>/
    SKILL.md        # required: YAML frontmatter + instructions
    scripts/        # optional: executable helpers
    references/     # optional: extra docs the skill can progressively disclose

Every skill folder is independently installable — directories like aiagentsdirectory.com/skills, skillsdirectory.com, and tools like openskills and prpm index this exact structure.

Skills in this repo

Skill	Purpose
aviation-regulations	Query aviation regulations, manuals, and publications via deepskyai.com's open search API (ICAO, FAA 14 CFR, EASA, CASA). No API key required.
flight-ops-calculators	Ten offline pilot calculators — gradient solver, Jet A-1 fuel uplift (3% rule), wind components, PA/DA, CAS/TAS/Mach, PET/PSR, TOD, W&B, ETOPS radius, holding entry. 34-test zero-dep suite.
approach-briefing	Generates a structured instrument-approach briefing from a plate (image or PDF). The hosting agent's native vision does the extraction; output includes a strict JSON schema and a pilot-facing briefing card. Optional scripts/deepsky_brief.py offloads to a deepsky server endpoint when one is live.

Installing a skill locally

Claude Code / Claude Desktop

Copy (or symlink) the skill folder into ~/.claude/skills/:

cp -R skills/aviation-regulations ~/.claude/skills/

Via openskills (any agent / IDE)

npx openskills install deepskyai/agent-tools -y

Via prpm

prpm install deepskyai/agent-tools

Authoring a new skill

1. Create skills/<your-skill>/SKILL.md with YAML frontmatter:

   ---
   name: your-skill
   description: One clear sentence describing when Claude should invoke this.
   ---
   

2. Keep SKILL.md short; push deep content into references/ for progressive disclosure.
3. Put runnable helpers in scripts/.

License

MIT — see LICENSE.