Skip to main content

Documentation Index

Fetch the complete documentation index at: https://docs.manticscore.com/llms.txt

Use this file to discover all available pages before exploring further.

The Search API lets you find projects and research using natural language rather than exact keyword matching. A query like “mobile banking for gig workers” will surface projects about freelancer payroll or earned wage access — not just ones with those exact words. Results respect your privacy settings — you only see your own projects plus any research others have explicitly made public. Search across projects and completed research by meaning. A query like “mobile banking for gig workers” will surface projects about freelancer payroll, contractor invoicing, and earned wage access — not just ones with those exact words.
​
q
string
required
Search query. Between 1 and 500 characters.
​
limit
number
default:"10"
Maximum results to return. Upper bound: 50.
curl "https://api.manticscore.com/projects/search?q=fintech+budgeting&limit=10" \
  -H "Authorization: Bearer <token>"
200 response
[
  {
    "id": "9f4e2a1b-...",
    "idea": "AI expense tracker for freelancers",
    "category": "fintech",
    "similarity": 0.9234,
    "is_own": true
  },
  {
    "id": "b7d3c1e9-...",
    "idea": "Smart budgeting app for contract workers",
    "category": "fintech",
    "similarity": 0.8811,
    "is_own": false
  }
]
​
id
string
required
UUID of the matching project or research job.
​
idea
string
required
The product idea text used to generate the embedding.
​
category
string
High-level category assigned during research (e.g. fintech, devtools). null if not yet categorized.
​
similarity
number
required
Cosine similarity score between 0 and 1. Higher values indicate a closer semantic match. Scores above 0.85 are typically strong matches.
​
is_own
boolean
required
true if the result belongs to the authenticated user. false means it is public research from another user.
Results include your own data plus research that others have set to reuse_scope: public. Private research from other users is never returned, regardless of similarity score.
Returns 503 if the semantic search service is temporarily unavailable. Retry after a moment.

​
Find similar projects

Given a specific project, find other projects and research with semantically similar ideas. This endpoint uses a pre-computed embedding for the project, so it is fast and returns results immediately.
​
limit
number
default:"5"
Maximum results to return. Upper bound: 20.
curl "https://api.manticscore.com/projects/9f4e2a1b-.../similar?limit=5" \
  -H "Authorization: Bearer <token>"
200 response
[
  {
    "id": "b7d3c1e9-...",
    "idea": "Smart budgeting app for contract workers",
    "category": "fintech",
    "similarity": 0.8811,
    "is_own": false
  }
]
The response shape is identical to the semantic search endpoint.
Returns 422 if the project does not have an embedding yet. Embeddings are generated asynchronously when research completes. Wait for the research pipeline to finish before calling this endpoint on a new project.

​
How semantic search works

ManticScore generates a semantic representation of each project idea text automatically when research completes. The search index includes:
  • Your own projects and research jobs (all visibility settings)
  • Other users’ research that has been set to reuse_scope: public
Search indexes are built asynchronously after each research run finishes. Until the index is ready, a project will not appear in search results or the similar endpoint — typically a few seconds after research completes.