| name | example-skill |
| description | Template skill — copy this folder to start a new quality-platform skill. Demonstrates scoring one FMEA Severity/Occurrence/Detection triple by calling the fmea_score MCP tool; do not install directly. |
Example skill — score one FMEA triple
This is the copy-from template for quality-platform skills. It shows the whole pattern in the
smallest possible case: the skill orchestrates, the engine decides. Read
skills/CONVENTIONS.md before you edit a copy of this folder.
When to use
The user supplies one Severity / Occurrence / Detection triple and wants its Risk Priority
Number and AIAG-VDA Action Priority. For a whole FMEA table, use the fmea_run tool instead.
Steps
- Connect to the quality-platform MCP server over stdio. The host normally has it configured
already; if not, it launches as
python -m mcp_app.server from the workspace root.
- Collect the three ratings from the user. Each is an integer on the AIAG 1-10 scale. Do not
invent a rating the user did not give, and do not translate ratings between scales — the
rating text lives in the FMEA app's own rating-scale tools (
fmea_list_scales,
fmea_get_scale), not in this skill.
- Call the
fmea_score tool with severity, occurrence and detection.
- Report the returned
rpn and action_priority verbatim. Never recompute, round,
re-derive or "sanity check" either value — the engine owns that arithmetic, and a second
opinion computed here is a defect, not a safeguard.
- If the tool returns an error (any rating outside 1-10), surface the error message to the
user and ask for a corrected rating. Do not retry with a clamped value.
scripts/call_fmea_score.py is the same flow as a runnable script, for when a shell call is
cheaper than a tool call.
Reference
