embeddings-query
Skill0 starsChanged 4 months agoMITView on GitHub
Query embeddings stored in a local Qdrant vector database. Supports retrieving a point by ID, semantic search (nearest-neighbor), and scrolling through a collection. Use after ingesting documents with the embeddings-ingest skill.
In this repository10 other files it gives its agents.
- arch-auditSkill
- check-pacman-updatesSkill
- embeddings-ingestSkill
- lease-hackrSkill
- perf-optimizationSkill
- pr-reviewSkill
- search-flightsSkill
- search-reposSkill
- security-auditSkill
- summarize-repo-changesSkill
The file
Files
Query Embeddings
Retrieve stored embeddings, search by semantic similarity, or scroll through a Qdrant collection.
Prerequisites
llama-embedding(from llama.cpp) — for creating query embeddings (search mode)curl— for Qdrant REST API callsjq— for formatting JSON output- Qdrant running locally (default:
http://localhost:6333)
Usage
scripts/query.sh -c <collection> [mode] [options]Required Arguments
| Flag | Description |
-c NAME | Qdrant collection name |
Modes (pick one)
| Flag | Description |
| (none) | Scroll — list points in the collection (paginated) |
-i ID | Get by ID — retrieve a specific point |
-q TEXT | Search — semantic nearest-neighbor search |
Optional Arguments
| Flag | Default | Description |
-n NUM | 5 | Number of results to return |
-o TOKEN | — | Scroll offset (use token from previous page) |
-u URL | http://localhost:6333 | Qdrant server URL |
--hf REPO | nomic-ai/nomic-embed-text-v1.5-GGUF:Q4_K_M | HuggingFace model for embedding queries |
-p POOL | mean | Pooling strategy |
-g NUM | auto | GPU layers |
--no-payload | off | Exclude payload from results |
--raw | off | Output raw JSON (no formatting) |
Examples
# Scroll — list first page of points
scripts/query.sh -c nomic-ai-nomic-embed-text-v1.5-gguf-q4-k-m
# Scroll with pagination
scripts/query.sh -c nomic-ai-nomic-embed-text-v1.5-gguf-q4-k-m -n 20
# Then next page with the offset token:
scripts/query.sh -c nomic-ai-nomic-embed-text-v1.5-gguf-q4-k-m -n 20 -o 2a3f1b4c
# Get a specific point by ID
scripts/query.sh -c nomic-ai-nomic-embed-text-v1.5-gguf-q4-k-m -i 2a3f1b4c
# Semantic search
scripts/query.sh -c nomic-ai-nomic-embed-text-v1.5-gguf-q4-k-m \
-q "What is the company revenue for 2024?"
# Search with more results
scripts/query.sh -c nomic-ai-nomic-embed-text-v1.5-gguf-q4-k-m \
-q "machine learning pipeline" -n 10Output Format
Results are displayed in a table:
# ID SOURCE CHUNK SCORE TEXT
──────────────────────────────────────────────────────────────────────────
1 a1b2c3d4 report.pdf 3 0.923 Q3 revenue increased by...
2 f5e6d7c8 report.pdf 2 0.891 Total revenue for 2024...
3 9g0h1i2j summary.txt 0 0.846 The company reported...Collection Names
Collection names are derived from the embedding model used during ingestion:
nomic-ai/nomic-embed-text-v1.5-GGUF:Q4_K_M→nomic-ai-nomic-embed-text-v1.5-gguf-q4-k-m
The same model must be used for searching as was used for ingestion to get meaningful results.
Does it work?
Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.
No reviews yet
People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.
- It worked
- Partly
- It didn't
Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.
Your trail
Files you open will collect here, so you can find them again.
Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.
Now do this for your own.
Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.