aliyun-dashvector-search

Name: aliyun-dashvector-search
Author: cinience

49installs34forks

How to install aliyun-dashvector-search?

cinience/alicloud-skills/aliyun-dashvector-search

$npx -y skills add cinience/alicloud-skills --skill aliyun-dashvector-search

Installs into the current project.

Summary

Use when building vector retrieval with DashVector using the Python SDK. Use when creating collections, upserting docs, and running similarity search with filters in Claude Code/Codex.

Files · 1

View on GitHub

SKILL.md

1	Category: provider
2
3	# DashVector Vector Search
4
5	Use DashVector to manage collections and perform vector similarity search with optional filters and sparse vectors.
6
7	## Prerequisites
8
9	- Install SDK (recommended in a venv to avoid PEP 668 limits):
10
11	```bash
12	python3 -m venv .venv
13	. .venv/bin/activate
14	python -m pip install dashvector
15	```
16	- Provide credentials and endpoint via environment variables:
17	- `DASHVECTOR_API_KEY`
18	- `DASHVECTOR_ENDPOINT` (cluster endpoint)
19
20	## Normalized operations
21
22	### Create collection
23	- `name` (str)
24	- `dimension` (int)
25	- `metric` (str: `cosine` \| `dotproduct` \| `euclidean`)
26	- `fields_schema` (optional dict of field types)
27
28	### Upsert docs
29	- `docs` list of `{id, vector, fields}` or tuples
30	- Supports `sparse_vector` and multi-vector collections
31
32	### Query docs
33	- `vector` or `id` (one required; if both empty, only filter is applied)
34	- `topk` (int)
35	- `filter` (SQL-like where clause)
36	- `output_fields` (list of field names)
37	- `include_vector` (bool)
38
39	## Quickstart (Python SDK)
40
41	```python
42	import os
43	import dashvector
44	from dashvector import Doc
45
46	client = dashvector.Client(
47	api_key=os.getenv("DASHVECTOR_API_KEY"),
48	endpoint=os.getenv("DASHVECTOR_ENDPOINT"),
49	)
50
51	# 1) Create a collection
52	ret = client.create(
53	name="docs",
54	dimension=768,
55	metric="cosine",
56	fields_schema={"title": str, "source": str, "chunk": int},
57	)
58	assert ret
59
60	# 2) Upsert docs
61	collection = client.get(name="docs")
62	ret = collection.upsert(
63	[
64	Doc(id="1", vector=[0.01] * 768, fields={"title": "Intro", "source": "kb", "chunk": 0}),
65	Doc(id="2", vector=[0.02] * 768, fields={"title": "FAQ", "source": "kb", "chunk": 1}),
66	]
67	)
68	assert ret
69
70	# 3) Query
71	ret = collection.query(
72	vector=[0.01] * 768,
73	topk=5,
74	filter="source = 'kb' AND chunk >= 0",
75	output_fields=["title", "source", "chunk"],
76	include_vector=False,
77	)
78	for doc in ret:
79	print(doc.id, doc.fields)
80	```
81
82	## Script quickstart
83
84	```bash
85	python skills/ai/search/aliyun-dashvector-search/scripts/quickstart.py
86	```
87
88	Environment variables:
89
90	- `DASHVECTOR_API_KEY`
91	- `DASHVECTOR_ENDPOINT`
92	- `DASHVECTOR_COLLECTION` (optional)
93	- `DASHVECTOR_DIMENSION` (optional)
94
95	Optional args: `--collection`, `--dimension`, `--topk`, `--filter`.
96
97	## Notes for Claude Code/Codex
98
99	- Prefer `upsert` for idempotent ingestion.
100	- Keep `dimension` aligned to your embedding model output size.
101	- Use filters to enforce tenant or dataset scoping.
102	- If using sparse vectors, pass `sparse_vector={token_id: weight, ...}` when upserting/querying.
103
104	## Error handling
105
106	- 401/403: invalid `DASHVECTOR_API_KEY`
107	- 400: invalid collection schema or dimension mismatch
108	- 429/5xx: retry with exponential backoff
109
110	## Validation
111
112	```bash
113	mkdir -p output/aliyun-dashvector-search
114	for f in skills/ai/search/aliyun-dashvector-search/scripts/*.py; do
115	python3 -m py_compile "$f"
116	done
117	echo "py_compile_ok" > output/aliyun-dashvector-search/validate.txt
118	```
119
120	Pass criteria: command exits 0 and `output/aliyun-dashvector-search/validate.txt` is generated.
121
122	## Output And Evidence
123
124	- Save artifacts, command outputs, and API response summaries under `output/aliyun-dashvector-search/`.
125	- Include key parameters (region/resource id/time range) in evidence files for reproducibility.
126
127	## Workflow
128
129	1) Confirm user intent, region, identifiers, and whether the operation is read-only or mutating.
130	2) Run one minimal read-only query first to verify connectivity and permissions.
131	3) Execute the target operation with explicit parameters and bounded scope.
132	4) Verify results and save output/evidence files.
133
134	## References
135
136	- DashVector Python SDK: `Client.create`, `Collection.upsert`, `Collection.query`
137
138	- Source list: `references/sources.md`

People who liked this also viewed

6 picks

Type

aliyun-dashvector-search

49installs34forks

View on GitHub

How to install aliyun-dashvector-search?

cinience/alicloud-skills/aliyun-dashvector-search

$npx -y skills add cinience/alicloud-skills --skill aliyun-dashvector-search

Installs into the current project.

Summary

Use when building vector retrieval with DashVector using the Python SDK. Use when creating collections, upserting docs, and running similarity search with filters in Claude Code/Codex.

Files · 1

View on GitHub

SKILL.md

1	Category: provider
2
3	# DashVector Vector Search
4
5	Use DashVector to manage collections and perform vector similarity search with optional filters and sparse vectors.
6
7	## Prerequisites
8
9	- Install SDK (recommended in a venv to avoid PEP 668 limits):
10
11	```bash
12	python3 -m venv .venv
13	. .venv/bin/activate
14	python -m pip install dashvector
15	```
16	- Provide credentials and endpoint via environment variables:
17	- `DASHVECTOR_API_KEY`
18	- `DASHVECTOR_ENDPOINT` (cluster endpoint)
19
20	## Normalized operations
21
22	### Create collection
23	- `name` (str)
24	- `dimension` (int)
25	- `metric` (str: `cosine` \| `dotproduct` \| `euclidean`)
26	- `fields_schema` (optional dict of field types)
27
28	### Upsert docs
29	- `docs` list of `{id, vector, fields}` or tuples
30	- Supports `sparse_vector` and multi-vector collections
31
32	### Query docs
33	- `vector` or `id` (one required; if both empty, only filter is applied)
34	- `topk` (int)
35	- `filter` (SQL-like where clause)
36	- `output_fields` (list of field names)
37	- `include_vector` (bool)
38
39	## Quickstart (Python SDK)
40
41	```python
42	import os
43	import dashvector
44	from dashvector import Doc
45
46	client = dashvector.Client(
47	api_key=os.getenv("DASHVECTOR_API_KEY"),
48	endpoint=os.getenv("DASHVECTOR_ENDPOINT"),
49	)
50
51	# 1) Create a collection
52	ret = client.create(
53	name="docs",
54	dimension=768,
55	metric="cosine",
56	fields_schema={"title": str, "source": str, "chunk": int},
57	)
58	assert ret
59
60	# 2) Upsert docs
61	collection = client.get(name="docs")
62	ret = collection.upsert(
63	[
64	Doc(id="1", vector=[0.01] * 768, fields={"title": "Intro", "source": "kb", "chunk": 0}),
65	Doc(id="2", vector=[0.02] * 768, fields={"title": "FAQ", "source": "kb", "chunk": 1}),
66	]
67	)
68	assert ret
69
70	# 3) Query
71	ret = collection.query(
72	vector=[0.01] * 768,
73	topk=5,
74	filter="source = 'kb' AND chunk >= 0",
75	output_fields=["title", "source", "chunk"],
76	include_vector=False,
77	)
78	for doc in ret:
79	print(doc.id, doc.fields)
80	```
81
82	## Script quickstart
83
84	```bash
85	python skills/ai/search/aliyun-dashvector-search/scripts/quickstart.py
86	```
87
88	Environment variables:
89
90	- `DASHVECTOR_API_KEY`
91	- `DASHVECTOR_ENDPOINT`
92	- `DASHVECTOR_COLLECTION` (optional)
93	- `DASHVECTOR_DIMENSION` (optional)
94
95	Optional args: `--collection`, `--dimension`, `--topk`, `--filter`.
96
97	## Notes for Claude Code/Codex
98
99	- Prefer `upsert` for idempotent ingestion.
100	- Keep `dimension` aligned to your embedding model output size.
101	- Use filters to enforce tenant or dataset scoping.
102	- If using sparse vectors, pass `sparse_vector={token_id: weight, ...}` when upserting/querying.
103
104	## Error handling
105
106	- 401/403: invalid `DASHVECTOR_API_KEY`
107	- 400: invalid collection schema or dimension mismatch
108	- 429/5xx: retry with exponential backoff
109
110	## Validation
111
112	```bash
113	mkdir -p output/aliyun-dashvector-search
114	for f in skills/ai/search/aliyun-dashvector-search/scripts/*.py; do
115	python3 -m py_compile "$f"
116	done
117	echo "py_compile_ok" > output/aliyun-dashvector-search/validate.txt
118	```
119
120	Pass criteria: command exits 0 and `output/aliyun-dashvector-search/validate.txt` is generated.
121
122	## Output And Evidence
123
124	- Save artifacts, command outputs, and API response summaries under `output/aliyun-dashvector-search/`.
125	- Include key parameters (region/resource id/time range) in evidence files for reproducibility.
126
127	## Workflow
128
129	1) Confirm user intent, region, identifiers, and whether the operation is read-only or mutating.
130	2) Run one minimal read-only query first to verify connectivity and permissions.
131	3) Execute the target operation with explicit parameters and bounded scope.
132	4) Verify results and save output/evidence files.
133
134	## References
135
136	- DashVector Python SDK: `Client.create`, `Collection.upsert`, `Collection.query`
137
138	- Source list: `references/sources.md`

People who liked this also viewed

6 picks

Type