The first official evaluation of the government bootcamp says the military activities like marching are "good" for the teenagers, but noted there was a lack of staff and Māori input.
At the same time, Oranga Tamariki said "alleged reoffending" had occurred.
The agency gave no details of the reoffending, such as if it was by one or more of the 10 boys after they had left the bootcamp youth justice facility, to move to the in-community phase late last year. The primary goal of the bootcamp was to cut re-offending.
RNZ is seeking more information.
The new evaluation report by Malatest International, noted weaknesses such as too few staff to give enough hours of individual clinical therapy, and too little te reo and te ao Māori, even though all the rangatahi were Māori and some were fluent.
But it concluded: "Despite the early stage of the MSA interventions and the complex needs of the cohort, there are indications of positive changes for many rangatahi" including the youth connecting with their whakapapa and each other in the group, and health conditions being addressed.
One person's feedback called it an "off the chart success" measured by zero physical fights in the three-month residential part, compared to much more fighting in facilities usually.
"They were motivated to leave residence as soon as they could so none of them wanted to fight," said the 89-page report.
One of the blackest marks in the evaluation was the clinical team felt their workloads were "unsustainable and felt under very high pressure throughout the pilot".
This was despite warnings from an NZDF-run early prototype of a boot camp, where overworked staff got to the point of feeling suicidal and suffered "serious mental harm", widely reported by media as the bootcamp was being designed.
At the pilot bootcamp, critical work sometimes fell on a single staff member, instead of two or more, the evaluation said.
It pushed for more clinical staff, a bigger "care team" and a fulltime administrator, though noted some staff had to live away from home for the pilot.
Oranga Tamariki said the evaluation was positive, and it talked about the "effectiveness of the residential phase".
The military elements included a "cohort approach" mirroring military training intakes, wearing uniforms, and cleaning up.
"The military aspects ... such as marching and drill were also very physical and a strong element of teamwork and coordination."
This was a "good fit" for the boys keen on fitness and sports, staff reported back, and "because it was absent from the lives they were living in the community when they offended".
The bootcamp began in July as a central but controversial part of the government's efforts to cut serious youth reoffending, which it is promising to follow up on this year with legislation that lets teens be sent to a camp - for the pilot, they volunteered - and be held up to nine months.
Under 'risks', the report mentions sunburn and someone getting a cut finger, but not that one of the boys died in a road crash and two absconded.
It did not give any scores or ratings, or make recommendations, such as definitively if more bootcamps should be held - favouring long lists of "learnings" instead - but it did state:
"Early indications support the ongoing refinement of military-style academy."
Oranga Tamariki said the teenagers seemed settled and engaged, "with many excelling educationally and achieving NCEA credits".
This was despite them spending less time on schoolwork than teens in a normal youth justice facility programme.
The multidisciplinary team of social workers, health, education and cultural professionals worked well, it said.
But staff shortages cropped up.
The bootcamp offered "far more clinical support" than usual, and it paid off: Rangatahi who had clammed up, by the end "can now sit down and tell you about what the good life model looks like to them", a staffer said.
But the small clinical team had a lot to do, and they also had to take over the transition-to-community plans "despite their existing workload challenges".
However, the evaluators said the clinical team had too much to do, cutting into the plans they could do for the children.
Oranga Tamariki noted it was "acutely aware" of the impact public scrutiny had on the group along with the death of a fellow participant.
"Our thoughts continue to be with this young person's whānau."
The design phase of the pilot was rushed, the evaluation found.
"There genuinely wasn't the time that you would like and would need to develop a really thoughtful, coherent programme," it quoted one staffer.
"Consultation with tangata whenua Rangitāne iwi began too late for it to be a significant driver... or for it to fulfil its role as tangata whenua" but later on there was more Māori input, though still not enough, the report said.
An local iwi community worker said: "We would have had veto over anything that we didn't see was going to be right for our city or for the young men."
Labour's children spokesperson, Willow-Jean Prime, said there were clear signs the bootcamp project was not going well.
"This is a 12-month pilot and we want to see the whole thing. We're already seeing clear signs this pilot isn't going well, and yet the minister is stonewalling anyone who digs deeper."
