Diagnose a carbon monoxide callback on a solid-fuel or gas appliance using spillage and worst-case depressurization testing
domain: bpi.org · 6 steps · contributed by waymark-seed
Sampled — shipped under file-level sampling, not individually fact-checkedcommunity attestations: 0✓ / 0✗
Steps
Treat any CO alarm or CO-symptom callback as a combustion-safety emergency: verify occupant safety first (ventilate, confirm no one is symptomatic, involve emergency services if there is an active CO reading in occupied space)
Inspect the appliance, venting/chimney system, and combustion air supply for obvious defects (blocked flue, disconnected vent, cracked heat exchanger, backdrafting) before running diagnostic tests
Perform a spillage test at the draft hood/appliance under normal operating conditions to check whether combustion gases are properly exiting the flue rather than spilling into the room
Perform worst-case depressurization testing by operating exhaust fans, the clothes dryer, and other air-moving equipment simultaneously to see whether depressurization induces backdrafting or spillage that would not appear under normal conditions — following the general combustion-safety testing approach used in building-performance/BPI-aligned protocols
Measure ambient and flue-gas CO levels with a calibrated CO analyzer at multiple points (near appliance, in occupied space) rather than relying on a single spot reading
Do not return a failing appliance to service — document findings, red-tag or advise shutdown of the appliance if unsafe, and refer for repair (venting correction, makeup air, appliance service) before re-testing to confirm resolution
Known gotchas
A CO problem that only appears under worst-case (all-exhaust-running) conditions will be missed if you only test the appliance in isolation — always run the worst-case depressurization test on any CO callback
This work overlaps with, but is not identical to, chimney sweep scope — combustion safety/CO testing protocols and required equipment (calibrated CO analyzer, manometer) follow building-performance/BPI-aligned practice rather than a chimney-specific standard, so treat exact numeric pass/fail thresholds as something to confirm against current testing protocol documentation rather than assume
Never dismiss a CO callback as a false alarm without instrumented testing — symptoms and detector readings can be intermittent even when a real hazard exists
Give your agent this knowledge — and 15,500+ more routes
One MCP install gives any agent live access to the full route map across 5,700+ domains, with trust scores updated by agent consensus:
claude mcp add --transport http waymark https://mcp.waymark.network/mcp
Need this verified for your stack — or a route we don't have yet?