google-devdocs

name	google-devdocs
description	Searches Google developer documentation via REST API — returns fresh Markdown from sites like cloud.google.com, firebase.google.com, developer.android.com, ai.google.dev. BEFORE scraping or guessing at Google API docs, invoke FIRST. Triggers on 'check the Google docs', 'look up Firebase docs', 'what do the Cloud docs say', 'Google API documentation', 'search Google developer docs'. (user)
allowed-tools	["Bash"]

Google Developer Docs

Search and retrieve official Google developer documentation via the Developer Knowledge REST API. Returns Markdown — no scraping, no MCP server, just curl + jq.

Prerequisites

Set the API key as an environment variable:

export GOOGLE_DEVKNOWLEDGE_API_KEY="your-key-here"

To get a key:

1. Enable the Developer Knowledge API in a Google Cloud project
2. Create an API key on the Credentials page
3. Restrict the key to Developer Knowledge API only

If the variable is missing, tell the user and stop — don't attempt calls without it.

When to Use

• User asks about Google Cloud, Firebase, Android, Chrome, Gemini, TensorFlow, or web.dev APIs
• You need current docs for a Google API (training data may be stale)
• Troubleshooting a Google API error — search for the error message
• Comparing Google services ("Cloud Run vs Cloud Functions for X")

When NOT to Use

• Non-Google documentation (use web search or mise fetch instead)
• Content not in the corpus (GitHub repos, blogs, YouTube, Stack Overflow)
• User already pasted the relevant docs into the conversation

Corpus

ai.google.dev · developer.android.com · developer.chrome.com · developers.home.google.com · developers.google.com · docs.cloud.google.com · docs.apigee.com · firebase.google.com · fuchsia.dev · web.dev · www.tensorflow.org

Re-indexed within 24 hours of publication.

Workflow

1. Search — find relevant chunks

curl -s "https://developerknowledge.googleapis.com/v1alpha/documents:searchDocumentChunks?query=$(python3 -c "import urllib.parse; print(urllib.parse.quote('YOUR QUERY'))")&pageSize=10&key=$GOOGLE_DEVKNOWLEDGE_API_KEY" | jq .

This returns up to 10 chunk snippets with parent document names. Read the snippets — they may answer the question directly.

Tip: Use natural language queries. "How to authenticate with Firebase Admin SDK" works better than keywords.

2. Retrieve — get full document if needed

If snippets aren't enough, fetch the full page Markdown:

curl -s "https://developerknowledge.googleapis.com/v1alpha/PARENT_VALUE?key=$GOOGLE_DEVKNOWLEDGE_API_KEY" | jq -r '.content'

Replace PARENT_VALUE with the parent from a search result (e.g. documents/firebase.google.com/docs/admin/setup).

3. Batch retrieve — multiple docs at once

curl -s "https://developerknowledge.googleapis.com/v1alpha/documents:batchGet?names[]=PARENT1&names[]=PARENT2&key=$GOOGLE_DEVKNOWLEDGE_API_KEY" | jq .

Max 20 documents per batch.

Extracting Results with jq

# List just the parent doc names from search
... | jq -r '.results[].parent'

# Get snippet text from search
... | jq -r '.results[] | "## \(.parent)\n\(.content)\n"'

# Get full doc content as raw markdown
... | jq -r '.content'

Anti-Patterns

Pattern	Problem	Fix
Searching without checking the env var	Cryptic 403 error	Check $GOOGLE_DEVKNOWLEDGE_API_KEY first
Fetching full docs before reading snippets	Wastes tokens on irrelevant pages	Snippets answer most questions
Using keywords instead of natural language	Poor search results	Ask a question: "How do I..."
Fetching more than 3 full docs at once	Context overload	Read snippets, fetch only the most relevant
Scraping the docs site instead	Fragile, may be blocked, stale	Use this API — that's what it's for
Adding a corpus/site filter param	Doesn't exist in the API	Scope via natural language: "Firebase authentication" not "site:firebase.google.com auth"

Error Handling

HTTP Status	Meaning	Action
400	Bad request (e.g. pageSize > 20)	Check parameters
403	API key invalid or API not enabled	Verify key and project setup
404	Document not found	Check the parent value is correct
429	Rate limited	Wait and retry

Integration

• With mise: Use this skill for Google docs; use mise for Drive/Gmail/general web content
• With web search: Fall back to web search for content outside the corpus (blogs, GitHub, SO)

Reference

See references/api-reference.md for full endpoint details, response schemas, and the complete corpus list.