Fallback And Resilience logo

Fallback And Resilience

Community
borski
fallback-and-resilience

What to do when a tool fails, an API hits a rate limit, or a site blocks scraping. Maps every primary tool to its best fallback so a single failure doesn't block the search.

Overview

Publisherborski
Repositorytravel-hacking-toolkit
Skill namefallback-and-resilience
Stars
690
Forks
75
Bundled files
Instructions only
LicenseMIT
Links
  • Markdown instructions

    A SKILL.md file the model loads on demand, so it only costs tokens when a request actually matches.

  • Works with any LLM

    AI skills are plain Markdown, not provider-specific code, so this works with GPT, Claude, Gemini, Grok, or a local model.

  • Self-contained

    Everything the model needs lives in the instructions — no extra files to sync.

  • Open source

    Published by borski on GitHub. Read the source before you install it.

Installation

Install the Fallback And Resilience AI skill in TypingMind to use it with any LLM, or drop it into another agent that reads SKILL.md.

1

Install in TypingMind

TypingMind installs a skill straight from its GitHub folder — it reads SKILL.md, bundles the resource files, and stores the result locally.

  1. Open the app and go to Plugins → Skills.
  2. Choose "Install from GitHub".
  3. Paste the skill folder URL below and confirm.
    https://github.com/borski/travel-hacking-toolkit/tree/main/plugins/travel-hacking-toolkit/skills/fallback-and-resilience
  4. Enable the skill in any chat where you want it available.
Plugins → Skills → Add skill → From GitHub URL, then paste the folder URL and press Continue.
2

Install in another agent

Any agent that reads the Agent Skills format can use this skill — copy the folder into that agent's skills directory.

Claude Code — .claude/skills
git clone --depth 1 https://github.com/borski/travel-hacking-toolkit.git /tmp/travel-hacking-toolkit
mkdir -p .claude/skills
cp -r /tmp/travel-hacking-toolkit/plugins/travel-hacking-toolkit/skills/fallback-and-resilience .claude/skills/fallback-and-resilience
Restart Claude Code after copying so it picks up the new skill.

Use it in TypingMind

Enable Fallback And Resilience in any TypingMind chat and the model takes it from there. Its name and description sit in the system prompt, and the moment a request matches, the model loads the full instructions itself — you never invoke it by hand, and it costs no tokens until it is actually used.

The model loads Fallback And Resilience on its own as soon as a request matches it.

Works with any AI model

AI skills are plain Markdown instructions rather than provider-specific code, so Fallback And Resilience is not tied to the model it was written for. Install it once in TypingMind and use it with GPT-5, Claude, Gemini, Grok, DeepSeek, Mistral, Llama, or a local model you run yourself — all on your own API keys.

  • Loaded only when it is needed

    The system prompt carries just the name and description. The instructions are fetched on the first matching request, so an idle skill costs nothing.

  • Switch models mid-chat

    Because the skill is instructions rather than code, changing model does not break it — the next model reads the same SKILL.md.

Skill instructions

This is the SKILL.md content the model loads. Read it before installing — a skill is instructions your model will follow.

Fallback and Resilience

Tools go down. APIs break. Have a backup plan for every search.

Primary ToolWhen It FailsFallback
DuffelAPI error or timeoutIgnav, Google Flights skill, Skiplagged
IgnavAPI errorDuffel, Google Flights skill, Skiplagged
Google Flightsagent-browser errorDuffel, Ignav, Skiplagged
Skiplagged502/timeout (Cloudflare issues)Kiwi.com MCP, Duffel, Ignav
Kiwi.comServer errorSkiplagged MCP, Duffel
Seats.aeroAPI error or stale dataCheck airline website directly, use Duffel for GDS inventory
SouthwestSW rate limiting or bot detectionWait a few minutes and retry. Use Docker (ghcr.io/borski/sw-fares) if running locally fails. Google Flights skill for SW cash prices as a fast fallback.
SerpAPIRate limit (100/mo free)Trivago for hotels, web search for destination discovery
TrivagoServer errorLiteAPI for hotels, SerpAPI Google Hotels
LiteAPIAuth error (401)Trivago MCP, SerpAPI Google Hotels
AirbnbScraping blockedSuggest user check airbnb.com directly
AwardWalletAPI errorAsk user for their balances directly
FerryhopperServer errorSerpAPI or web search for ferry routes
Atlas ObscuraScript errorWeb search for "unusual things to do in [destination]"
Chase TravelLogin failure or CSRF issuesUse Duffel/Ignav for cash prices. Note that Points Boost and Edit detection are Chase-only.
Amex TravelLogin failure or form changesUse Duffel/Ignav for cash prices. Note that IAP fares and FHR/THC detection are Amex-only.
Deutsche Bahn (db-vendo)Library error or bahn.de outageSerpAPI Google Travel Explore, web search for "DB train [origin] [dest]", or fall back to confirming flight options instead.
Wikipedia airportsAPI rate limit (rare) or page missingWeb search "[airport name] airlines destinations" returns the same data via Google.
Fare tool says "no results"Could be no availability OR no serviceUse wikipedia-airports to confirm whether the route is flown at all. If Wikipedia lists the destination as served, fare tools have no availability on that date. If Wikipedia doesn't list it, the route doesn't exist. Saves you from asking the user to retry searches that can never succeed.

General Rules

  • If an MCP server returns an error, try the curl-based skill equivalent (or vice versa).
  • If a paid API hits its rate limit, switch to a free alternative.
  • Never give up after one tool fails. Always try at least one fallback.
  • Tell the user which source you used. "Skiplagged was down, so I checked Kiwi.com instead."

Never Claim MCP Is Unavailable Without Calling It

Do not declare an MCP server "not available" from a glance at your tool list — attempt an actual tool call first. The Skiplagged and Kiwi servers need no API keys and are wired into every supported harness config; a session launched from the repo directory has them. If a call genuinely fails, quote the error to the user and then fall back.

And when falling back, never silently substitute ad-hoc browser scraping for the toolkit's sources. Improvised scraping (e.g., driving a browser to Google Flights by hand) produces prices that miss most of the market. If you're ever reduced to it, say so explicitly and label every price from it as low-confidence: "MCP sources were unreachable, so this is a single scraped quote — treat it as a ceiling, not a market price."

IP Allowlist Errors (Common First-Time Failure)

Several APIs in the toolkit enforce IP allowlists on the developer key. If you see one of these error signatures, do NOT rabbit-hole on auth or quotas. Suggest the user whitelist their IP first.

APIError SignatureWhitelist Page
AwardWallet{"error": "access_denied", "code": "IP_DENIED"}https://business.awardwallet.com/profile/api
TripAdvisorUser is not authorized to access this resource with an explicit denyhttps://www.tripadvisor.com/developers

Diagnostic flow when an API call fails with an auth-shaped error on first use:

  1. Run curl ifconfig.me to get the user's current outbound IP.
  2. Check if the failure signature matches an IP allowlist (table above or similar phrasing like "explicit deny" / "IP_DENIED" / "not whitelisted").
  3. Tell the user: "This looks like an IP allowlist issue, not a key issue. Add <their IP> at , wait 1-5 min for AWS edge cache propagation, then retry."
  4. While they whitelist, gracefully degrade: continue without that data source, noting it's pending. Don't block the rest of the workflow.
  5. Multi-IP gotcha: residential CGNAT, VPN exits, hotel wifi, and travel locations all need separate entries. If they hit the same error from a new network, the IP changed.

"No Cached Availability" Is Not the Final Word

When Seats.aero returns no results for a route + program combination, that means Seats.aero has not scraped it recently. It does NOT mean the award is unbookable. When a reachable program shows no cached results, search the airline's website directly before declaring awards dead.

Patchright-Based Skills (Southwest, AA, Chase, Amex, TaW)

These hit websites directly with an undetected browser. Common failure modes:

  • Login form changed. The site updated its DOM. Selectors break. Update the skill.
  • 2FA loop. Some skills (AA, Amex) handle email 2FA. Make sure 2FA delivery method is set correctly in the account.
  • Bot detection. If you get "unusual activity" or CAPTCHA pages, the persistent profile may be flagged. Try a fresh profile or wait an hour.
  • Headless detection. Patchright runs headed. If running locally fails, use the Docker image (xvfb provides virtual display).

Docker Image Failures

If docker pull ghcr.io/borski/... fails:

  • unauthorized: Run docker logout ghcr.io then retry. Or login with a GitHub PAT that has read:packages. Or build locally with docker build -t <tag> skills/<skill>/.
  • Network timeout: Retry with --platform linux/amd64 if you're on ARM and getting checksum mismatches.

Frequently asked questions

What does the Fallback And Resilience AI skill do?

What to do when a tool fails, an API hits a rate limit, or a site blocks scraping. Maps every primary tool to its best fallback so a single failure doesn't block the search.

Why use Fallback And Resilience on TypingMind?

Because you install it once and use it with any model. Fallback And Resilience is plain Markdown rather than provider-specific code, so the same skill runs on GPT-5, Claude, Gemini, Grok, or a local model — and you can switch model mid-chat without it breaking. TypingMind runs on your own API keys, so you pay providers directly instead of a per-seat subscription, and your skills and chats stay in your own storage.

How do I install Fallback And Resilience in TypingMind?

Open Plugins → Skills → Install from GitHub in TypingMind and paste https://github.com/borski/travel-hacking-toolkit/tree/main/plugins/travel-hacking-toolkit/skills/fallback-and-resilience. TypingMind reads its SKILL.md and installs it as a skill you can enable per chat.

Which AI models can use Fallback And Resilience?

Any model you connect in TypingMind. AI skills are plain Markdown instructions rather than provider-specific code, so GPT, Claude, Gemini, Grok, and local models can all load this skill when a request matches it.

How many AI models can I use with Fallback And Resilience?

As many as you like. As long as a model supports skills, you can use Fallback And Resilience with it — GPT, Claude, Gemini, Grok, DeepSeek, Mistral, Llama and more — all on TypingMind with your own API keys.

Is the Fallback And Resilience AI skill free?

Yes. It is published on GitHub by borski under the MIT license. You only pay your own AI provider for the tokens you use.

What are AI skills?

An AI skill is a reusable instruction bundle that teaches an AI model how to do one specific task. It follows the open Agent Skills format: a SKILL.md file with a name and description, plus any scripts, templates or reference files the model may need. The model reads the instructions only when your request matches the skill, so an installed skill costs nothing until it is used.

How are AI skills different from plugins or MCP servers?

A plugin or MCP server gives a model new tools to call — code that runs somewhere and returns a result. An AI skill gives the model knowledge and process instead: how to approach a task, which steps to follow, what good output looks like. Skills are plain Markdown, so they need no server, no API key and no runtime, and they work with any model.

View all

Set up your own AI workspace now

Get notified about new features and future giveaways by subscribing to our newsletter 👇