The security contract hearing took place the following week. On the table were the agreement, six months of incident logs, resident complaints, pilot metrics and the veterinary report. Maua did not ask for Silas to be fired. She asked for the definition of safety to change.
The committee treasurer began with cost. Rono Secure was the second-lowest bidder. The next provider would cost twelve percent more.
Silas said, "Our property-incident record is strong. The zero-stray program reduced complaints."
"It may have reduced stray complaints," Maua said. "That is not identical to child-safety performance."
She presented measurable alternatives: named handoff completion, backup activation time, gate-change notification, missing-child escalation timeline.
"Security cannot be responsible for parents," Silas said.
"It is not. Parents own the receiver. School owns release. Security owns gate operation, communication and response. We need the interfaces."
That distinction mattered.
They examined the logs. Twenty-six animal entries. No category for *child separated from receiver*. Two complaints about early gate closure had been recorded as "resident inconvenience."
"Why?"
"The categories are fixed," Silas said.
"Then the data cannot reveal a risk that has no category."
The committee lawyer asked about the 18:05 preliminary report. Silas admitted the form required a probable cause before the incident had been resolved.
Dr. Wanjiku's report did not support animal causation. The police timeline documented storm, route disruption, shelter and rescue. The committee ordered an independent review of the reporting process.
Silas pushed back on paperwork. "My staff will fill forms instead of guarding."
Joseph, invited as a shift representative, spoke. "We want clear rules too. That day I was told to close at 4:20 and allow escort exceptions. There was no escort log. If I let a child pass, I can be blamed. If I stop the child, I can also be blamed."
The room went quiet.
They negotiated minimum clauses: - zero-stray remains an animal-control objective, not the master safety KPI; - gate changes affecting the school corridor must reach the school route desk; - child-handoff incident response time is measured; - guard exceptions receive a minimal log; - preliminary incident reports separate verified fact, working hypothesis and unverified cause; - foster and community animal care follow veterinary and welfare guidance.
Then the lawyer asked each clause to have an owner. Security supervisor. School release desk. Parent. Route coordinator. Named foster caregiver. Independent reviewer.
The economics changed too. Zero-stray had carried a five-percent performance bonus. The committee split the incentive between animal-control response, gate communication compliance, and incident-response documentation.
Silas did the numbers on a calculator. "Possible. Guard training will cost extra."
The committee approved a one-time training budget in exchange for the revised report process and a six-month renewal instead of a full year.
Neither side loved the compromise.
Before the vote, the committee tested one scenario: another storm tomorrow. School holds the child until a receiver is confirmed. Security sends a gate-change notice. Route desk changes the marked path. Parent activates backup. Police escalation begins if a child misses the timed checkpoint.
Silas asked, "Where is Kito in this scenario?"
Maua answered, "Wherever his foster caregiver has him. The system does not need to know."
That sentence ended the confusion more cleanly than any speech.
The committee approved the revised contract and opened review of the premature report.
Afterward Mama Tesa handed Maua her notebook.
"This isn't mine," Maua said.
"I know. Copy the route pages for training. The original stays with me. Remove the children's names."
The routes could become shared knowledge without turning children's private history into training material.
The contract had changed.
And Mama Tesa's notebook was no longer the secret safety system of one child.