Debugging

When an intent does something unexpected — no rules matched, a branch took the wrong path, state did not update, or an event never appeared — you need more than logs. Causet records a forensics trail for every intent so you can inspect, fork, replay, and repair.

Practical path after a retrofit failure:

Inspect the timeline
Walk a failed workflow
Fork → replay → repair projection / notify again

In this section

Page	What it covers
Timeline	Inspect event history and rule steps
Ledger Events	What committed — patches, emits, intent status
Entity Inspector	Browse entities, replay scrubber, diff, rebuild
Replay	Simulate failure, fork, replay, repair
Forensics CLI	`causet forensics` — drift check and per-decision repair

Three views of the same execution

View	Question it answers
Timeline	How did rules evaluate?
Ledger events	What changed at commit time?
Intent status	Did it succeed? Committed, rejected, or failed

All three share intent ID. For a given entity, committed routes and ledger entries also share cursor.

Typical workflow

Find the intent or entity — Entities list, Decision Timeline, or intent ID.
Check outcome — committed, rejected, or failed?
Read the decision route — which rule blocks ran?
Verify the ledger commit — patch and emits as expected?
Inspect state — live snapshot or replay at cursor.

Control plane tools

Tool	Nav location	Purpose
Decision Timeline	Operate → Decision Logs	Intent/rule timeline, forensics, replay & branch
Entity Inspector	Operate → Entities	Snapshots, scrubber, diff, rebuild
Runs	Operate → Runs	Cross-intent execution monitoring (links to both above)

Filter by stream, entity, fork, intent type, and cursor or time range.

Decision routes vs AI Decisions

Term	Meaning
Decision routes	Runtime forensics for all rule steps
AI Decisions	Product DSL LLM calls (`op: decision`)

AI calls appear as one step in the decision route. Authoring: AI. Reading the trace: Timeline.