1version: v1beta1
2kind: Spicepod
3name: shadcn-ui
4
5runtime:
6  dataset_load_parallelism: 1
7
8datasets:
9  - from: github:github.com/shadcn-ui/ui/files/main
10    name: shadcn_ui.docs
11    description: Shadcn UI documentation from https://ui.shadcn.com/docs
12    params:
13      file_format: md
14      github_client_id: ${secrets:GITHUB_CLIENT_ID}
15      github_private_key: ${secrets:GITHUB_PRIVATE_KEY}
16      github_installation_id: ${secrets:GITHUB_INSTALLATION_ID}
17      include: "apps/www/content/docs/**/*.mdx"
18    acceleration:
19      enabled: true
20      refresh_check_interval: 4h
21      refresh_jitter_enabled: true
22      refresh_jitter_max: 15m
23    columns:
24      - name: content
25        embeddings:
26          - from: openai_embeddings
27            row_id:
28              - path
29            chunking:
30              enabled: true
31              target_chunk_size: 512
32              overlap_size: 128
33              trim_whitespace: true
34
35  - from: github:github.com/shadcn-ui/ui/files/main
36    name: shadcn_ui.components
37    description: Shadcn UI default components source code
38    params:
39      github_client_id: ${secrets:GITHUB_CLIENT_ID}
40      github_private_key: ${secrets:GITHUB_PRIVATE_KEY}
41      github_installation_id: ${secrets:GITHUB_INSTALLATION_ID}
42      include: "apps/www/registry/default/ui/**/*.tsx"
43    acceleration:
44      enabled: true
45      refresh_check_interval: 4h
46      refresh_jitter_enabled: true
47      refresh_jitter_max: 15m
48    columns:
49      - name: content
50        embeddings:
51          - from: openai_embeddings
52            row_id:
53              - path
54            chunking:
55              enabled: true
56              target_chunk_size: 512
57              overlap_size: 128
58              trim_whitespace: true
59
60  - from: github:github.com/shadcn-ui/ui/issues
61    name: shadcn_ui.issues
62    description: Shadcn UI GitHub issues from github.com/shadcn-ui/ui
63    time_column: created_at
64    params:
65      file_format: md
66      github_client_id: ${secrets:GITHUB_CLIENT_ID}
67      github_private_key: ${secrets:GITHUB_PRIVATE_KEY}
68      github_installation_id: ${secrets:GITHUB_INSTALLATION_ID}
69    acceleration:
70      enabled: true
71      refresh_mode: full
72      refresh_check_interval: 4h
73      refresh_jitter_enabled: true
74      refresh_jitter_max: 15m
75      refresh_sql: "SELECT * FROM shadcn_ui.issues LIMIT 100"
76    # embeddings:
77    #   - column: title
78    #     column_pk:
79    #       - id
80    #     use: openai_embeddings
81
82embeddings:
83  - from: openai
84    name: openai_embeddings
85    params:
86      openai_api_key: ${ secrets:OPENAI_API_KEY }
87

1version: v1beta1
2kind: Spicepod
3name: shadcn-ui
4
5runtime:
6  dataset_load_parallelism: 1
7
8datasets:
9  - from: github:github.com/shadcn-ui/ui/files/main
10    name: shadcn_ui.docs
11    description: Shadcn UI documentation from https://ui.shadcn.com/docs
12    params:
13      file_format: md
14      github_client_id: ${secrets:GITHUB_CLIENT_ID}
15      github_private_key: ${secrets:GITHUB_PRIVATE_KEY}
16      github_installation_id: ${secrets:GITHUB_INSTALLATION_ID}
17      include: "apps/www/content/docs/**/*.mdx"
18    acceleration:
19      enabled: true
20      refresh_check_interval: 4h
21      refresh_jitter_enabled: true
22      refresh_jitter_max: 15m
23    columns:
24      - name: content
25        embeddings:
26          - from: openai_embeddings
27            row_id:
28              - path
29            chunking:
30              enabled: true
31              target_chunk_size: 512
32              overlap_size: 128
33              trim_whitespace: true
34
35  - from: github:github.com/shadcn-ui/ui/files/main
36    name: shadcn_ui.components
37    description: Shadcn UI default components source code
38    params:
39      github_client_id: ${secrets:GITHUB_CLIENT_ID}
40      github_private_key: ${secrets:GITHUB_PRIVATE_KEY}
41      github_installation_id: ${secrets:GITHUB_INSTALLATION_ID}
42      include: "apps/www/registry/default/ui/**/*.tsx"
43    acceleration:
44      enabled: true
45      refresh_check_interval: 4h
46      refresh_jitter_enabled: true
47      refresh_jitter_max: 15m
48    columns:
49      - name: content
50        embeddings:
51          - from: openai_embeddings
52            row_id:
53              - path
54            chunking:
55              enabled: true
56              target_chunk_size: 512
57              overlap_size: 128
58              trim_whitespace: true
59
60  - from: github:github.com/shadcn-ui/ui/issues
61    name: shadcn_ui.issues
62    description: Shadcn UI GitHub issues from github.com/shadcn-ui/ui
63    time_column: created_at
64    params:
65      file_format: md
66      github_client_id: ${secrets:GITHUB_CLIENT_ID}
67      github_private_key: ${secrets:GITHUB_PRIVATE_KEY}
68      github_installation_id: ${secrets:GITHUB_INSTALLATION_ID}
69    acceleration:
70      enabled: true
71      refresh_mode: full
72      refresh_check_interval: 4h
73      refresh_jitter_enabled: true
74      refresh_jitter_max: 15m
75      refresh_sql: "SELECT * FROM shadcn_ui.issues LIMIT 100"
76    # embeddings:
77    #   - column: title
78    #     column_pk:
79    #       - id
80    #     use: openai_embeddings
81
82embeddings:
83  - from: openai
84    name: openai_embeddings
85    params:
86      openai_api_key: ${ secrets:OPENAI_API_KEY }
87