Skip to main content
Without with, this is the edge read: the rows of tckg.read_edges. With with=strength, each row also carries strength: one item per measurement fact that points at the edge (see write measurements), valued the way resolve values a fact. The server computes nothing; it masks the measurements by as_of and picks with the fact’s policy. An edge nobody measured has strength: [], and so does every edge that is not causal (SUPPORTS, SUPERSEDES, MENTIONS and the rest are never scored). That is the normal answer, not an error.

​
Query parameters

​
timestamp or date
required
Knowledge time. Edges, their endpoints and the measurements must have been known by then. A date is widened to its end of day (UTC). Without it the request is a 400 (TCKG_AS_OF_REQUIRED).
​
timestamp
Content time. Only edges in force at valid_at (default as_of). A measurement whose sample starts after valid_at answers no_value_at.
​
string
Edges whose source is in this space, and only measurements written in this space.
​
string
causal, argumentative, temporal or general.
​
string
strength adds the strength column. Any other value is a 400.
​
integer
With with=strength: keep only edges with an answered item of at least this grade. Without it no edge is filtered.
​
string
With with=strength: keep only the items of this horizon. Edges are not filtered; with min_grade, the grade is judged on this horizon’s items.
​
timestamp
With with=strength: apply the fact declarations in force at this instant, as in resolve. A measurement backfilled to before its fact was declared is undeclared_fact unless you pass it.
min_grade, horizon and rules_as_of without with=strength are a 400 (TCKG_STRENGTH_WITHOUT_WITH).

​
Response

​
EdgeRow[]
source_id, target_id, edge_type, family, tenant, status, confidence, lag, mechanism, properties, asserted_at, captured_at, valid, known_at, backfill_batch, and with with=strength, strength. The first fifteen are the same with or without with.
​
object[]
One item per measurement fact, in fact_key order:
  • fact_key, horizon
  • status, reason, value: the fact’s resolved value, or no_answer and why: unresolved_conflict (two measurements known at the same instant), no_value_at (valid_at is outside the sample), undeclared_fact (no declaration then; see rules_as_of), edge_mismatch (the winning measurement was taken on another edge)
  • id, known_at: the winning measurement and when it became known
  • candidates: how many measurements were considered
  • prereg: {commitment_id, known_at, before_measurement} when the measurement links a preregistration; before_measurement says the commitment was known before the measurement
​
object
Scoped to space. masked counts what was hidden because it was known after as_of, measurements included.
Ask the same question with as_of before a measurement was written and the edge reads strength: [] again: a measurement is known from when the server recorded it, never earlier.