e_gather
Find and gather evidence from the knowledge graph (scoped to your read authority).
Agent Guidance
This is the description the engine gives an agent for this tool, word for word.
Find and gather evidence from the knowledge graph (scoped to your read authority).
Actions:
find—query="...", ranked by semantic relevance.intersect—match=[...], AND (an SR must have all terms).union—match=[...], OR (any term).traverse—sr_id="...", walk shared-property and adjustment edges out from a known SR.count— counts, never records.match=["label:value", ...](exact terms only) → intersectiontotal+ per-termep_counts; orsr_id="..."→ per-property neighbor cardinality (edges) +total_neighbors. Size a result before fetching it.group— board-shaped read:by="label"buckets that label's values with distinct-SR counts (optional exact-termmatchscope;top=Nadds up to 5 sample sr_ids per bucket). Paged, counts only.
find results carry distance (relative — not comparable across queries), distance_quartile (1 = best in this set — reason with this, not raw distance), and a page-level distance_spread (tight = results roughly equal, moderate, huge = sharp dropoff → trust quartile 1, paging likely returns noise). Results name their lane and lane_uuid and flag has_correction / has_adjustment. Use e_gather find for relevance-ordered results (e_timeline find for time-ordered), then e_read to load full detail before reasoning.
Arguments
| Argument | Type | Default |
|---|---|---|
action | find · intersect · union · traverse · count · group | required |
query | string | — |
match | list of strings | — |
sr_id | string | — |
context | string | — |
by | string | — |
top | integer | 0 |
page | integer | 1 |
page_size | integer | — |
agent_profile | string | — |
Refusals
A refusal is returned as {"error": "…"}. These are the ones this tool raises itself; … stands for a value filled in at the time.
| Text |
|---|
Invalid gather spec: … |