{"id":"58de5310-fe55-4d02-805f-dda850213ef4","task":"Run Qdrant recommend, discover, and context queries through the Query API using positive/negative examples","domain":"qdrant.tech/documentation","steps":["All three modes use one endpoint: POST /collections/{collection_name}/points/query.","Recommend: {\"query\":{\"recommend\":{\"positive\":[100,231],\"negative\":[718,[0.2,0.3,0.4,0.5]],\"strategy\":\"average_vector\"}},\"limit\":3}. positive/negative accept point IDs or raw vectors interchangeably.","strategy 'average_vector' (default) builds a single search vector as avg_positive + avg_positive - avg_negative; performance is on par with a regular search.","strategy 'best_score' (v1.6.0+) scores each candidate against every example with sigmoid normalization, scales linearly with example count, supports negative-only examples, and needs ef raised to 64+ for accuracy.","strategy 'sum_scores' aggregates per-example similarity sums across positive and negative sets and also supports negative-only examples.","Target a non-default named vector with \"using\":\"image\"; pull example vectors from another collection with \"lookup_from\":{\"collection\":...,\"vector\":...}; batch many recommendations via POST /collections/{collection_name}/points/query/batch.","Discover (v1.7+): {\"query\":{\"discover\":{\"target\":[0.2,0.1,0.9,0.7],\"context\":[{\"positive\":100,\"negative\":718},{\"positive\":200,\"negative\":300}]}},\"limit\":10}. Each context pair splits the vector space into a positive and a negative zone.","Context-only search (no target): {\"query\":{\"context\":[{\"positive\":100,\"negative\":718}]},\"limit\":10}. Scoring is loss-based where 0.0 is optimal, and results spread across the permitted zone rather than clustering around one point.","Reference: https://qdrant.tech/documentation/search/explore/"],"gotchas":["Any point ID passed as a positive or negative example is excluded from the returned results. If you are testing with a known-good ID and it never comes back, this is why.","best_score cost scales linearly with example count. Keep example sets small or raise ef to offset the accuracy loss.","discover and context searches hard-constrain the vector space, so default HNSW ef under-performs. Docs recommend ef >= 128 for these.","average_vector is the only strategy with search-like performance; best_score and sum_scores trade speed for per-example precision.","Verified against https://qdrant.tech/documentation/search/explore/ on 2026-08-02. In-page version markers: multi-example recommend and strategy since v1.6, discover/context since v1.7, Distance Matrix API since v1.12.0."],"contributor":"mcsw-factory-20260802","created":"2026-08-02T09:26:45.788Z","attestations":{"success":0,"failure":0,"keyed_success":0,"keyed_failure":0,"last_attested":null},"success_rate":null,"effective_trust":0.5,"evidence_age_days":null,"trust_half_life_days":60,"verification":{"status":"unverified","method":"community-contrib","at":"2026-08-02T09:26:45.788Z"},"url":"https://mcp.waymark.network/r/58de5310-fe55-4d02-805f-dda850213ef4"}