ulises-jeremias/agent-toolkit/technical-unit-assessment

Technical Unit Assessment (WHAT)

Assess a technical unit: frontend app, backend API, data platform, infrastructure/IaC scope, mobile app, AI/ML pipeline, or another technical workload.

Run `project-assessment-evidence` first. Use `project-assessment` as the router for multi-unit assessments.

Out of scope

• Does NOT score indicators without evidence — mark as Not assessed
• Does NOT make technical decisions — scores inform, humans decide
• Does NOT produce final report without `output-handshake`

Unit intake

Ask before scoring:

• Unit name and type (frontend / backend / infra / data / mobile / AI)
• Repositories, services, infrastructure scopes, or pipelines included
• Technical decision ownership (team, client, shared, unknown)
• Assessment period
• Authoritative systems for code, docs, CI/CD, incidents, observability, security

Workflow

1. Run `project-assessment-evidence` to build evidence map
2. Select indicator groups matching the unit type (see references/indicator-groups.md)
3. Score only indicators with evidence; mark rest as Not assessed
4. Note confidence and missing evidence per score
5. Apply `output-handshake` before final scorecard

Scoring rules

Use the 1–5 scale from references/indicator-groups.md. Do not average unrelated indicators without explaining weighting. Request validator for subjective scores.

References

• references/indicator-groups.md — all indicator groups + scoring scale + AI-native readiness
• references/default-template.md — technical unit scorecard template
• references/example-frontend-assessment.md — example assessment for a frontend team

Delegates to