UnisonTimeline
Read an entity's fact history over time. asOf queries answer what the brain believed on a given date — useful for debugging decisions made with stale information.
Every fact in Unison is bitemporal: it records when the fact became true (validFrom/validTo) and when the brain learned it (recordedAt). The timeline API surfaces this history so you can answer "what did we believe on a given date" — not just "what do we believe now."
Read an entity's timeline
from unisonlabs import UnisonBrain
client = UnisonBrain()
events = client.entities.timeline("<id>")unison timeline <id>Response: an ordered list of fact events — recorded, corrected, invalidated — each with timestamps and the fact payload.
Filter by date range
unison timeline <id> --from 2026-05-01 --to 2026-06-01asOf — time-travel reads
asOf on facts answers "what was believed at that moment." Useful for debugging a decision that used stale memory:
# Facts as of May 1 (via search with asOf)
curl 'https://brain.unisonlabs.ai/v1/brain/search?q=deploy+process&asOf=2026-05-01T00:00:00Z' \
-H "Authorization: Bearer $UNISON_TOKEN"Context recall always answers from current memory — asOf is not supported on /v1/brain/context.
Why bitemporality matters
Two common failure modes it prevents:
- Silent overwrite: when two agents record contradictory facts, both are preserved with timestamps. The contradiction is visible and resolvable — not silently last-write-wins.
- Phantom knowledge: a fact that was true in March but corrected in April should not influence a recall in May.
asOflets you reconstruct exactly what the brain knew at decision time, so post-mortems have ground truth.
How corrections appear in the timeline
When client.facts.correct() records a correction, the timeline shows:
2026-03-10 recorded depends_on = "Postgres 16" confidence: 0.8
2026-05-10 corrected depends_on = "Postgres 17.2" confidence: 0.95
↳ supersedes fact_abcThe old fact is not deleted — it is accessible via asOf=2026-04-01.
See also: Entities & Facts · Knowledge graph concept · API reference
Entities & Facts
The knowledge graph: resolve an entity, read its facts, record new facts. Corrections supersede rather than overwrite — no silent last-write-wins.
Scopes & visibility
Where you write is who sees it. /private/ for the caller, /workspace/teams/<slug>/ for a team, /workspace/ for the whole workspace. No permission matrix.