Exception Raise
Purpose
Deterministic ranking of store-launch readiness exceptions: weekly promo revenue at stake × launch proximity × price-conflict factor (references/execution-rules.md #5.1) — impact, not count. The engine ranks; this skill quotes.
When to use
After readiness-check whenever anything in checked.json is missing or conflicted.
When NOT to Use
- Gaps in product content — missing images, bullet copy, GTINs, attributes for a marketplace or channel listing — use the catalog enrichment completeness-score instead, or action-register instead for the fix order. This skill only knows store-launch readiness.
- "Are we ready" / "any gaps" (the test itself) — use readiness-check instead.
- Drafting the readiness checklist, shift brief or DM escalation text — use brief-draft instead.
- The pack or store records still need assembling — use pack-ingest instead for the pack, use store-map instead for the store records.
Where the data comes from
This plugin ships no sample data. Every input is the user's own Microsoft 365 content:
- Files the user names, links or attaches in the chat.
- Their OneDrive or SharePoint, found with Work IQ search (Microsoft 365 search) by campaign ID, store ID or file name when no location is given.
- Email attachments (for example a store manager's note) found the same way.
Read each source with the file reader (ReadFileContent) and cite it by file name and row or page. Write the working JSON files (velocity.json, ranked.json) to the session working folder, never into the plugin. If a required source cannot be found, say which one is missing and ask the user where it is. Never invent it, estimate it or fall back to example values.
Inputs
checked.json (the {readiness} hop, from readiness-check or supplied by the user) + the sales velocity extract from the user's Microsoft 365 content (see Where the data comes from). Convert the extract to velocity.json, citing its rows, in exactly this shape: { "<SKU>": {"weekly_units": <n>, "promo_price": <price>} }. Never estimate weekly units. Rules: references/execution-rules.md #5.1.
Steps
- Run the engine:
python scripts/exception_rank.py --checked checked.json --velocity velocity.json --out ranked.json - Quote the ranking verbatim with the numbers. A small missing signage kit on the top-velocity SKU outranks five gaps on slow movers — say so with the revenue figures.
- Flag the proximity factor (2× inside 3 days of launch) and the price-conflict factor (1.5×) from
provenance.ranking_factors; top exceptions inside 3 days go to the district manager. - Validate the output with the contract validator tool:
python tools/validate_contract.py ranked.json.
python scripts/exception_rank.py --checked checked.json --velocity velocity.json --out ranked.json
# exception_rank: 4 exception(s), top=PRICE-GTIN-2002, proximity=2.0x, price_conflict=1.5x -> ranked.json
Output format
ranked.json — the {exceptions} hop. Report as a markdown table in engine order: # | req_id | status | detail | revenue at stake / week | rank score | citation; then one sentence on impact-vs-count with the figures; then the district-manager list.
Grounding requirements
Every exception cites its pack line (or price-file row) and the velocity extract.
Guardrails
- Ranking is engine-only; never reorder, promote or demote an exception because a gap "looks small" or someone would rather the district manager did not see it.
- Never drop or hide a PRICE CONFLICT — it stays in the list and in
escalations[]until HQ resolves the price file. - Never estimate weekly units; if the velocity extract is missing, the engine fails closed — list the exceptions unranked and ask for the extract.
- Do not draft the brief here — hand
ranked.jsonto brief-draft. - Always show the ranked table for review before escalation goes anywhere.
Escalation / uncertainty
Top exceptions within 3 days of launch carry the 2× proximity factor and go to the district manager. If readiness.blocked is true (#6), say the ranking is provisional until the missing input arrives.