Aliyun Milvus Search

Use when working with AliCloud Milvus (serverless) with PyMilvus to create collections, insert vectors, and run filtered similarity search. Optimized for Cla...

Security Scan

The skill appears to implement exactly what it says (connect to AliCloud Milvus via PyMilvus), but there is a transparency/mapping problem: it requires Milvus credentials via environment variables even though the registry metadata declares none — this mismatch and the need to provide credentials warrants caution.

Purpose & Capability

The code and SKILL.md match the stated purpose: they use PyMilvus to create collections, insert vectors, and run searches against an AliCloud Milvus endpoint. The requested artifacts (MILVUS_URI, MILVUS_TOKEN, MILVUS_DB, optional collection/dimension env vars) are appropriate for that functionality. However, the registry metadata lists no required environment variables even though the runtime instructions and script require credentials — a transparency mismatch.

Instruction Scope

SKILL.md and scripts stay within the declared scope: they instruct installing pymilvus, reading MILVUS_* env vars, performing a minimal read call, then running create/insert/search operations and saving local output under output/aliyun-milvus-search. There are no instructions to read unrelated system files, phone home to unknown endpoints, or exfiltrate data to third-party services.

Install Mechanism

This is an instruction-only skill with a small helper script. It recommends installing pymilvus via pip (in a virtualenv) — a standard, low-risk approach. There is no bundled download from arbitrary URLs or other installation of external binaries.

Credentials

The runtime requires MILVUS_URI and MILVUS_TOKEN (credentials) and optionally MILVUS_DB, MILVUS_COLLECTION, MILVUS_DIMENSION. Those variables are proportional to the stated purpose. The concern is that the skill registry declares no required environment variables or primary credential — meaning the manifest underreports the credential requirements. Supplying MILVUS_TOKEN grants access to your Milvus instance; ensure you do not provide production/global credentials and prefer scoped or temporary credentials.

Persistence & Privilege

The skill does not request always:true and does not modify other skills or system-wide settings. It writes outputs locally under an output/ directory which is normal. Note: the agent is allowed autonomous invocation by default (disable-model-invocation is false). Combined with credential access this increases blast radius, but autonomous invocation alone is platform default and not a sole reason for rejection.

What to consider before installing

This skill appears to do what it claims (connect and operate on AliCloud Milvus via PyMilvus), but pay attention before supplying credentials. The registry metadata does not list the environment variables the script actually requires (MILVUS_URI and MILVUS_TOKEN), so: 1) do not run this with long-lived or production credentials — create scoped or temporary credentials for testing; 2) run first in an isolated environment (local VM or test account) and in a virtualenv as recommended; 3) verify network access (VPC or public endpoint) and that the token has least privilege required for the operations you want; 4) check the pymilvus version and audit the quickstart.py source if you need stricter assurance; 5) consider whether you want the agent to be allowed to invoke this autonomously while having access to environment variables — if not, avoid giving the agent persistent environment access or disable autonomous invocation for this skill. Finally, ask the publisher (or registry) to correct the manifest to explicitly declare required env vars so you can make an informed decision.

Category: provider

AliCloud Milvus (Serverless) via PyMilvus

This skill uses standard PyMilvus APIs to connect to AliCloud Milvus and run vector search.

Prerequisites

• Install SDK (recommended in a venv to avoid PEP 668 limits):

python3 -m venv .venv
. .venv/bin/activate
python -m pip install --upgrade pymilvus

• Provide connection via environment variables:
  • MILVUS_URI (e.g. http://<host>:19530)
  • MILVUS_TOKEN (<username>:<password>)
  • MILVUS_DB (default: default)

Quickstart (Python)

import os
from pymilvus import MilvusClient

client = MilvusClient(
    uri=os.getenv("MILVUS_URI"),
    token=os.getenv("MILVUS_TOKEN"),
    db_name=os.getenv("MILVUS_DB", "default"),
)

# 1) Create a collection
client.create_collection(
    collection_name="docs",
    dimension=768,
)

# 2) Insert data
items = [
    {"id": 1, "vector": [0.01] * 768, "source": "kb", "chunk": 0},
    {"id": 2, "vector": [0.02] * 768, "source": "kb", "chunk": 1},
]
client.insert(collection_name="docs", data=items)

# 3) Search
query_vectors = [[0.01] * 768]
res = client.search(
    collection_name="docs",
    data=query_vectors,
    limit=5,
    filter='source == "kb" and chunk >= 0',
    output_fields=["source", "chunk"],
)
print(res)

Script quickstart

python skills/ai/search/aliyun-milvus-search/scripts/quickstart.py

Environment variables:

• MILVUS_URI
• MILVUS_TOKEN
• MILVUS_DB (optional)
• MILVUS_COLLECTION (optional)
• MILVUS_DIMENSION (optional)

Optional args: --collection, --dimension, --limit, --filter.

Notes for Claude Code/Codex

• Insert is async; wait a few seconds before searching newly inserted data.
• Keep vector dimension aligned with your embedding model.
• Use filters to enforce tenant scoping or dataset partitions.

Error handling

• Auth errors: check MILVUS_TOKEN and instance permissions.
• Dimension mismatch: ensure all vectors match collection dimension.
• Network errors: verify VPC/public access settings on the instance.

Validation

mkdir -p output/aliyun-milvus-search
for f in skills/ai/search/aliyun-milvus-search/scripts/*.py; do
  python3 -m py_compile "$f"
done
echo "py_compile_ok" > output/aliyun-milvus-search/validate.txt

Pass criteria: command exits 0 and output/aliyun-milvus-search/validate.txt is generated.

Output And Evidence

• Save artifacts, command outputs, and API response summaries under output/aliyun-milvus-search/.
• Include key parameters (region/resource id/time range) in evidence files for reproducibility.

Workflow

1. Confirm user intent, region, identifiers, and whether the operation is read-only or mutating.
2. Run one minimal read-only query first to verify connectivity and permissions.
3. Execute the target operation with explicit parameters and bounded scope.
4. Verify results and save output/evidence files.

References

• PyMilvus MilvusClient examples for AliCloud Milvus

• Source list: references/sources.md