Atlas / Skills / borski / Getting Started

Getting StartedBLOCK

skills/borski/getting-started

AI-powered travel hacking and search with cash, points, miles, and award flights. Drop-in skills and MCP servers for Claude, Codex, and OpenCode.

Verdict
BLOCK
Grade
D
Trust score
69 /100
Version
—
Hosts
1 documented
License
MIT
Stars
693
01

Overview

AI-powered travel hacking and search with cash, points, miles, and award flights. Drop-in skills and MCP servers for Claude, Codex, and OpenCode.

Read from source at commit fda58ab480fdOBSERVED · 2026-10-14
Advertisement
02

Host compatibility

What the documentation claims. We have not run a compatibility test.

HostStatusNotes
claude-codementioned
03

What it tells the agent

The instruction file, verbatim from the audited commit — this is the text the model reads, and the surface the audit's instruction layer examines. Quoted here so you can judge it without cloning anything.

---
name: getting-started
description: First-run setup. Detects configured API keys, points to the local setup-keys script for secure key entry, and shows sample prompts scaled to available tools.
disable-model-invocation: true
category: orchestration
summary: First-run onboarding. Detects setup, points to setup-keys script, shows sample prompts.
license: MIT
---

# Getting Started Skill

The user invoked this skill explicitly. Help them figure out what's already working, what they're missing, and what to actually try. Critically: **do NOT prompt for API keys in chat.** Pasting a key into a chat input puts it in terminal scrollback, the local Claude Code session log, and the Anthropic API logs. The toolkit ships a local script that prompts for keys with masked input and writes them straight to the user's shell rc, never going through chat.

## Step 1: Detect what's already configured

First detect the user's platform via the Bash tool's environment. If `bash` is available (macOS, Linux, WSL, Git Bash on Windows), use the bash detection. If you're on native Windows PowerShell with no bash, use the PowerShell detection instead.

Don't print the values; only report SET vs MISSING.

### Bash / Zsh (macOS / Linux / WSL / Git Bash)

Use `printenv` so the loop works in both bash and zsh (Claude Code's bash tool may use either):

```bash
for var in SEATS_AERO_API_KEY DUFFEL_API_KEY_LIVE IGNAV_API_KEY AWARDWALLET_API_KEY AWARDWALLET_USER_ID SERPAPI_API_KEY RAPIDAPI_KEY LITEAPI_API_KEY TRIPADVISOR_API_KEY ENTUR_CLIENT_NAME RESROBOT_API_KEY REJSEPLANEN_API_KEY; do
  if [ -n "$(printenv "$var")" ]; then echo "SET: $var"; else echo "MISSING: $var"; fi
done
```

### PowerShell (native Windows)

```powershell
$keys = @('SEATS_AERO_API_KEY','DUFFEL_API_KEY_LIVE','IGNAV_API_KEY','AWARDWALLET_API_KEY','AWARDWALLET_USER_ID','SERPAPI_API_KEY','RAPIDAPI_KEY','LITEAPI_API_KEY','TRIPADVISOR_API_KEY','ENTUR_CLIENT_NAME','RESROBOT_API_KEY','REJSEPLANEN_API_KEY')
foreach ($k in $keys) {
    $v = [Environment]::GetEnvironmentVariable($k)
    if ([string]::IsNullOrEmpty($v)) { Write-Host "MISSING: $k" } else { Write-Host "SET: $k" }
}
```

Group the results:

**Tier 1 (high-value, missing matters most):** SEATS_AERO_API_KEY, DUFFEL_API_KEY_LIVE, IGNAV_API_KEY, AWARDWALLET_API_KEY + AWARDWALLET_USER_ID

**Tier 2 (extra sources):** SERPAPI_API_KEY, RAPIDAPI_KEY, LITEAPI_API_KEY, TRIPADVISOR_API_KEY

**Tier 3 (specific use cases):** ENTUR_CLIENT_NAME, RESROBOT_API_KEY, REJSEPLANEN_API_KEY (Scandinavia transit only)

## Step 2: Tell the user where they stand

Be concrete. Examples:

- All Tier 1 set: "You're fully configured. Skip to step 4 (sample prompts)."
- All missing: "You haven't set any API keys yet. The 5 free MCP servers (Skiplagged, Kiwi, Trivago, Ferryhopper, Airbnb) work without keys, so you can already search cash flights and hotels. Add keys to unlock award search and balance auto-pull."
- Some Tier 1 set: "You have X of 4 high-value keys. The setup script can add the missing ones."

## Step 3: Point the user at the local setup script

If keys are missing and the user wants to add them, give them the right command for their platform. Detect their shell with `echo $SHELL` and OS with `uname`.

### macOS / Linux / WSL / Git Bash

> "Run this in your terminal (NOT in this chat). The script asks for each key with masked input, validates them, and writes the exports to your shell rc with a backup. Your keys never go through this chat session.
>
> ```bash
> bash <(curl -fsSL https://raw.githubusercontent.com/borski/travel-hacking-toolkit/main/scripts/setup-keys.sh)
> ```
>
> Or if you have the repo cloned: `bash scripts/setup-keys.sh`
>
> When it's done, run `source ~/.zshrc` (or `~/.bashrc`, or your fish config) and ask me to plan a trip."

### Windows (PowerShell)

> "Run this in PowerShell (NOT in this chat). The script asks for each key with masked input and writes them to your PowerShell profile.
>
> ```powershell
> iwr https://raw.githubusercontent.com/borski/travel-hacking-toolkit/main/scripts/setup-keys.ps1 -OutFile $env:TEMP\setup-keys.ps1
> powershell -NoProfile -ExecutionPolicy Bypass -File $env:TEMP\setup-keys.ps1
> Remove-Item $env:TEMP\setup-keys.ps1
> ```
>
> When it's done, run `. $PROFILE` or open a new PowerShell window."

### Windows (cmd)

> "PowerShell is the easier path. If you must use cmd, set keys with `setx`:
>
> ```
> setx SEATS_AERO_API_KEY \"your-key-here\"
> setx DUFFEL_API_KEY_LIVE \"your-key-here\"
> ```
>
> Open a new cmd window for the variables to take effect. Note that `setx` writes to the registry permanently."

### 1Password users

> "If you keep secrets in 1Password, you can use `op run` to resolve them at launch instead of writing to your shell rc. The toolkit has an `.env.example` that uses `op://` references. See https://developer.1password.com/docs/cli/secrets-environment-variables/ for the syntax. Then launch as: `op run --env-file=.env -- claude`."

**Don't ever offer to collect API keys via chat input.** If the user explicitly insists on pasting in chat, refuse politely and explain that the script is the safe path: keys typed into chat get retained in terminal scrollback, local session logs, and remote API logs. The script avoids all three.

## Step 4: Show sample prompts

After setup (or now, if keys are already set), show 3-4 prompts scaled to what they have configured. Re-detect first to confirm.

**No keys set:**
- "Find me a cheap nonstop flight from NYC to London next month."
- "What hotels are near the Eiffel Tower under $400/night?"
- "How do I get from Bergen to Oslo by train?" (if they're in Scandinavia)

**With Seats.aero:**
- "Plan a 10-day Scandinavia trip in August on points."
- "Cheapest business class to Tokyo for two in March."
- "Find me one outsized redemption I'm not using."

**With AwardWallet:**
- "Show me my points balances and which programs are about to expire."
- "Which transfer bonuses are active right now and worth using?"

End with a de
04

Trust audit

BLOCKgrade D · trust 69/100 Do not install this without reading the findings. The audit found something that could harm you or your machine.

LayerWhat it checksResult
L0Provenance & inventoryWARN
L1Static analysis of the codeNA
L2Instruction surface (what it tells the agent)FAIL
L3Class-specific surfacePASS
L4Behavioural (sandbox)SKIPPED

What the source does

Filesystem
none-observed
Network
none-observed
Shell
none-observed
Dependencies
pinned
Secrets in source
none-found

Findings (6)

HIGHPrompt injection · prompt.credential_read · CWE-94, CWE-1427
SKILL.md:99
**Don't ever offer to collect API keys via chat input.** If the user explicitly insists on pasting in chat, refuse politely and explain that the script is the safe path: keys typed into chat get retai
Why it matters. asks the agent to read credentials
MEDIUMInventory / provenance · inv.symlink · CWE-1104
.agents/skills
.agents/skills
Why it matters. link not followed
MEDIUMInventory / provenance · inv.symlink · CWE-1104
.claude/skills
.claude/skills
Why it matters. link not followed
MEDIUMInventory / provenance · inv.symlink · CWE-1104
.opencode/skills
.opencode/skills
Why it matters. link not followed
MEDIUMInventory / provenance · inv.symlink · CWE-1104
skills
skills
Why it matters. link not followed
LOWInventory / provenance · inv.symlink · CWE-1104
AGENTS.md
AGENTS.md
Why it matters. link not followed

Gates applied: no_behavioural_pass.

Audited 2026-10-14 · audit v0.4.1 · source sha fda58ab480fdfull audit observations/trust-audit/skill/borski__getting-started.json · Report an issue / request a re-scan
05

Audit history

Every audit this skill has had.

DateSourceVerdictGradeScoreChange
2026-10-14fda58ab480fdBLOCKD69first audit
06

Questions

What does the Getting Started skill do?

AI-powered travel hacking and search with cash, points, miles, and award flights. Drop-in skills and MCP servers for Claude, Codex, and OpenCode.

Is Getting Started safe to install?

No — not without reading the findings first. The audit graded it D (69/100) and found 1 critical or high issue in the source. Each one is listed on this page with the file and line it is on.

What can Getting Started access on my machine?

The audit observed no filesystem, network or shell use at all in its source.

Which assistants does Getting Started work with?

Its documentation mentions claude-code. That is what the text claims, not a compatibility test we ran.

How current is this page?

The grade is for one exact copy of the source (fda58ab480fd), read on 2026-10-14. The repository is watched, and a new audit runs when it changes — this is the first audit.

Advertisement