Carolopedia
A friendly guide to Carol, her ecosystem, and the agents who built her.
📖About & Usage
About
Reviewing the initiative — judging the whole initiative once its steps are done, inferring the reviewing transition and grading bypass work against its success criteria.
Where it fits
This is one stage of the Build Initiatives service. The owner and the agents who run it are listed under the team below, and the other blocks of the service are linked at the bottom of this page.
🛠️Team & droids
Elrond Block owner
At initiative review, Elrond decides whether the whole initiative is actually done, and he carries the two droids that make and infer that call. The Initiative Review droid reads each success criterion and its current status, counts review cycles used against the cap, and decides automatically: all criteria met means close, all failing means continue — and on mixed results it asks Claude whether the unmet criteria are essential to the initiative's purpose or acceptable gaps, escalating only the genuinely ambiguous cases (mixed results plus tight budget plus uncertainty plus essential failures) to Orion, and recording the decision and any gaps. The Initiative Review Inferer infers the reviewing transition itself: when a plan step flips to done it re-reads the committed status, counts the steps in that phase, and if every step is done and no review row exists yet for the phase, flips the initiative to reviewing — it does not review anything itself; it only sets up the flip the Manager reads to trigger the reviewer. This matters because it is the gate between work-finished and initiative-closed, and it has to distinguish essential failures from cosmetic ones rather than blindly passing or failing. It fires when a phase's steps all complete (the inferer) and then on the reviewing initiative (the review droid), with the escalate path reserved for the truly uncertain calls.
Reviewing the initiative is where finished work is judged as a whole, and Albus's role here is the planner-independent grader for bypass work, which the normal reviewer can't touch. The Albus Twin Reviewer grades Albus bypass initiatives from Albus's own bypass book plus the initiatives DB — explicitly never the planner — serving as the Albus-side twin of Elrond's reviewer under the three-book model. This matters because bypass work skips the normal planner-driven path, so a planner-based reviewer would have nothing to grade it against; without this twin, bypassed initiatives would have no grading authority at all and could close ungraded. It fires once a bypass initiative's work is done and it enters review, drawing its success-criteria evidence from the bypass book rather than planner artifacts. In the normal case it produces a grade for the bypass initiative against its own recorded criteria, keeping bypass work held to a standard even though it took the planner-skipping lane.