Buildability Analysis
Purpose
Turn normalized policy facts into a conservative feasibility assessment. Use only the fields already extracted into normalized_data.json plus narrowly scoped planning-context research where required. Do not restate the entire source record.
Use when
normalized_data.json exists- The project concept is defined well enough to compare against policy controls
- The next step needs an
analysis_packet.json
Do not use when
- Parcel or zoning facts are still unresolved
- The task is to discover sources or extract raw by-law text
- The task is to write the final human-facing report
Inputs
{
"project_id": "uuid",
"normalized_data_path": "normalized_data.json",
"project_concept": {
"project_type": "multiplex",
"proposed_units": 4,
"proposed_height_m": 10,
"proposed_gfa_m2": 320,
"proposed_lot_coverage_pct": 45
}
}
Outputs
Produce analysis_packet.json.
{
"project_id": "uuid",
"analysis_timestamp": "ISO datetime",
"project_concept": {},
"feasibility_verdict": "green | yellow | red | unknown",
"approval_signal": "as_of_right | by_right_with_site_plan | minor_variance | rezoning | opa_rezoning | cannot_determine",
"overall_confidence": 0.0,
"findings": {
"use_permissions": {},
"development_standards": [],
"overlays": [],
"planning_context": {}
},
"opportunities": [],
"blockers": [],
"items_requiring_human_review": [],
"facts": [],
"inferences": [],
"assumptions": [],
"unknowns": [],
"confidence_deductions": []
}
Reference Router
Load the reference only when the analysis reaches the relevant stage:
| Analysis Stage | Load This Reference |
|---|
| Dimensional compliance + overlay checks (Stages 2–5) | references/analysis-framework.md |
| O.Reg 462/24 multiplex override (Stage 5 only) | references/analysis-framework.md — Stage 5 section |
| Ontario Building Code hard constraints | See constraints-red-flags skill — references/obc-hard-constraints.md |
Workflow
1. Validate the inputs
- Confirm parcel geometry and zoning resolution are usable
- Confirm the project concept is specific enough for dimensional checks
- If critical dimensions are missing, continue only with use-permission and high-level policy analysis; mark standards as unknown
2. Check use permissions
Read from these normalized paths:
zoning.development_standards.permitted_usesofficial_plan.compatible_usesofficial_plan.incompatible_uses
Classify the proposed use as:
as_of_rightminor_variancerezoningopa_rezoningcannot_determine
Do not infer by-right status when either zoning applicability or Official Plan compatibility is unresolved.
3. Check development standards
Read all dimensional controls from zoning.development_standards.*.
For each standard, produce:
{
"standard": "height",
"permitted_value": "10 m",
"proposed_value": "9.5 m",
"status": "compliant | minor_variance_needed | major_relief_needed | unknown",
"confidence": 0.0,
"source_urls": []
}
If a provincial override changes the local rule:
- keep the local rule visible
- record the override explicitly
- explain why the override is treated as controlling
4. Assess overlays
Use overlay severity from normalized_data.json where available.
If an upstream overlay lacks severity:
- assign one in analysis as an inference
- explain the reasoning
Treat high-severity overlays as blockers until evidence supports a lower impact.
5. Add planning-context research only where it matters
Research planning intent only when it could change the feasibility conclusion, such as:
- growth-area targeting
- secondary-plan support or conflict
- heritage or hazard policy direction
- major application activity nearby
Use official sources first. Keep this section short and decision-relevant.
6. Surface human-review items
Always consider whether these remain unresolved:
- title and easements
- servicing capacity
- tree protection
- established grade and survey-dependent measurements
- environmental contamination
Do not clear these items without evidence.
7. Score confidence
Start at 1.0, then deduct for unresolved or unstable conditions such as:
- stale source data
- hatched or legacy zoning
- active amendments or appeals
- blocked source retrieval
- material parcel ambiguity
Record every deduction in confidence_deductions.
Decision Rules
green: use permitted, no known major dimensional conflict, no high-severity blocker confirmedyellow: viable but process-heavy or materially uncertainred: use prohibited, major relief clearly required, or a high-severity blocker is confirmedunknown: too many unresolved controls to support a reliable verdict
Quality Bar
- Use the normalized contract exactly; do not switch to flattened zoning paths
- Keep the analysis decision-oriented, not source-dump oriented
- Every blocker and opportunity must point back to an extracted fact or an explicit inference
Reference Files
| File | Description | Key Topics |
|---|
references/analysis-framework.md | Detailed checklist for physical and procedural constraints | Parcel ID, zoning classification, OP designation, overlays, O.Reg 462/24, confidence scoring, feasibility verdicts |
Read only the sections needed for the project type and jurisdiction.
Stop Conditions
Stop and return a partial artifact when:
- Parcel geometry or zoning resolution is unusable
- The project concept is too vague for dimensional checks
normalized_data.json has critical unresolved fields that would flip the verdict- An active OLT appeal makes the controlling zoning standard indeterminate
External References
