Back to catalog / Agent workflows
Analyze protected document collections in chunks
Fetch large unstructured collections through an authenticated proxy, process bounded batches in memory, and answer cross-file questions without first moving documents into a vector database.
Goal
Fetch large unstructured collections through an authenticated proxy, process bounded batches in memory, and answer cross-file questions without first moving documents into a vector database.
Steps
- Configure the internal file proxy with Secret Manager authentication and an approved test corpus. [src-google-analyze-protected-document-collections-in-chunks]
- Fetch files in bounded parallel batches, summarize/analyze each batch, and record coverage counts. [src-google-analyze-protected-document-collections-in-chunks]
- Synthesize across batches while retaining supporting file IDs and missing items. [src-google-analyze-protected-document-collections-in-chunks]
How to verify
Use known facts from distant files to check the answer and compare expected versus processed file counts.
Troubleshooting
If coverage is short, inspect proxy auth, pagination cursor, and failed batches; reduce batch size when context or memory limits are reached.
Permissions
Prerequisites
- Google ADK/Gemini environment, internal proxy, Secret Manager access, and approved test files; no vector database dependency.
Permissions
- Reads private documents and sends text to Gemini API; require authorization and retention review and never write downloaded files to public storage.
Limitations
Throughput and context are bounded; state coverage and do not treat unfetched files as evidence of absence.
Pinned upstream source
- Repository
- google/adk-samples
- File path
- contrib/python/high-volume-document-analyzer/README.md
- Pinned commit
9252494a7aaafca8fe4d81a04f15392345a5779b- Source checked
- 2026-10-14 · Reviewed against the pinned upstream source
Repository metrics
These values describe the upstream repository, not popularity or likes for an individual skill or playbook.
- Stars
- Not collected
- Forks
- Not collected
- Observed at
- Not collected
No metrics snapshot is available yet.
License: Apache-2.0
Main topic
Turn ecommerce campaign materials into a reviewable weekly brief