Search

Shoehorn provides full-text search across your entire platform, powered by Meilisearch. Results appear instantly as you type with typo tolerance, synonym expansion, and relevance ranking.

Using Search

Click the search bar or press / from any page. Results stream in as you type.

What is Searchable

Search covers 9 content types:

Type	Searchable fields (priority order)
Entities	name, description, tags, metadata
Documentation	title, search terms, description, content, tags
Users	name, username, email, department
Teams	name, description, type
Repositories	name, full name, description, language, topics
Workflows	name, slug, description, tags
Molds	name, slug, description, language, tags
Pipelines	name, description, provider
Doc Sites	title, description, URL

Fields listed first have higher relevance weight. A match on an entity name ranks higher than a match in its metadata.

Search Features

Typo tolerance

Search finds results even with misspellings. Words with 4+ characters tolerate 1 typo, words with 8+ characters tolerate 2 typos.

Examples:

• paymnt finds payment
• kuberntes finds kubernetes
• deploymnt finds deployment

Synonyms

Built-in synonyms expand your queries automatically:

You type	Also matches
k8s	kubernetes, kube
svc	service
deploy	deployment
repo	repository
db	database
app	application
config	configuration
docker	dockerhub
git	github, gitlab

Additional synonyms are registered per content type. For example, in repositories: gh matches github, gl matches gitlab, bb matches bitbucket.

Ranking

Results are ranked by:

1. Word match, more matched query terms rank higher
2. Typo count, fewer typos rank higher
3. Proximity, matched terms closer together rank higher
4. Field priority, match in name ranks above match in description
5. Sort, user-specified sort order (if any)
6. Exactness, exact matches rank above prefix matches
7. Recency, newer items get a boost

Filtering

Narrow results by type, lifecycle, owner, or tags using the filter panel or query parameters.

Semantic search (documentation)

Documentation search supports hybrid keyword + semantic matching. When you ask a question-like query (starting with “how”, “what”, “why”, etc.) or use 5+ words, the search engine combines keyword relevance with meaning-based vector similarity for better results.

Search API

curl "https://shoehorn.example.com/api/v1/search?q=payment&type=service" \
  -H "Authorization: Bearer <token>"

Query parameters

Parameter	Description	Example
q	Search query	payment
type	Filter by content type	entity, doc, user, team, repository, workflow, mold, pipeline
owner	Filter by team	payments-team
lifecycle	Filter by lifecycle	production
tags	Filter by tags	critical
limit	Results per page (default 20)	20
offset	Pagination offset	0
sort_by	Sort field	relevance, date, title, type

Autocomplete

curl "https://shoehorn.example.com/api/v1/search/autocomplete?q=pay&limit=5" \
  -H "Authorization: Bearer <token>"

Autocomplete combines:

• Query completions
• Content-based suggestions
• Your recent search history
• Trending searches across the platform

Results are cached per user for 5 minutes.

Indexing

Entities are indexed automatically when:

• Created via API, manifest, K8s agent, or Terraform
• Updated (metadata, annotations, or enrichment changes)
• Enriched (README fetched, scorecard evaluated, entityFile applied)

The search index rebuilds automatically if it becomes stale. You can check indexing status at:

GET /api/v1/indexing/status

Access Control

Search results are filtered by your tenant and permissions. You only see items you have access to. The search engine enforces:

• Tenant isolation (you only see your tenant’s data)
• Public/private visibility
• Team-based access controls
• Soft-delete exclusion (deleted items are hidden)