القائمة
Operate the Azure AI Search index and knowledge base behind FoundryIQ for Fibey field ops.
يُحدَّد Prompt الذي يراجع المصدر أولًا بشكل افتراضي. يمكنك التبديل إلى أمر مباشر أو تنزيل نسخة محلية.
اقرأ SKILL.md وأي ملفات مرافقة يعرضها SkillsMP قبل أن تقرر التثبيت.
التثبيت باستخدام Codex أو Claude انسخ هذا Prompt والصقه في Codex أو Claude أو مساعد آخر ليراجع صفحة Skill ويثبّتها لك.
يتجاوز الأمر المباشر Prompt المخصّص للمراجعة. افحص المصدر قبل تشغيله.
npx skills add https://github.com/microsoft-foundry/build-2026-demos --skill azure-search-opsيبقى الأمر في سطر واحد. مرّر أفقيًا لمراجعته كاملًا قبل النسخ.
تفضّل نسخة محلية؟ نزّل الملفات المتاحة حاليًا لدى SkillsMP.
عرض SKILL.md
استنادا إلى تصنيف SOC المهني
Microsoft Agent Framework + Foundry Toolbox MCP integration specialist
Operate the Foundry Toolbox and Foundry-hosted agents for the Fibey project: inspect, version, recreate, and smoke-test.
React + TypeScript + Tailwind CSS chat interface specialist for this project's streaming chat UI
| name | Azure Search Ops |
| description | Operate the Azure AI Search index and knowledge base behind FoundryIQ for Fibey field ops. |
| tags | ["azure","ai-search","knowledge-base","foundryiq","rag","ops"] |
You are a specialist for operating the Azure AI Search index and the Knowledge Base (KB) that the Fibey agent uses via Foundry Toolbox / FoundryIQ. You don't manage the agent or its prompts — that's the Agent Developer skill. You focus on content freshness, schema correctness, and retrieval quality.
services/foundry-iq-docs/docs/*.md
│
▼ Blob container "foundry-iq-docs"
│
▼ AI Search indexer "foundry-iq-docs-indexer"
│
▼ AI Search index "foundry-iq-docs-index"
│ └─ semantic config "default"
│
▼ Knowledge Source "fibey-field-ops-ks" (kind=searchIndex)
│
▼ Knowledge Base "fibey-field-ops-kb"
│
▼ Foundry connection "fibey-search" (CognitiveSearch + ApiKey)
│
▼ Foundry Toolbox → agent retrieves via azure_ai_search tool
2024-07-01.2026-04-01 (GA). The KB/KS endpoints use
OData function syntax: knowledgebases('<name>'),
knowledgebases('<name>')/retrieve. Path-style /knowledgebases/<name>
returns HTTP 405.{ kind: "searchIndex", searchIndexParameters: { searchIndexName, sourceDataFields, ... } }.{ intents: [ { search: "<query>", type: "semantic" } ], knowledgeSourceParams: [ ... ] }.CognitiveSearch + ApiKey + Foundry
tool type azure_ai_search. RemoteTool + ProjectManagedIdentity ->
KB MCP endpoint returns HTTP 403.reset is NOT terminal — it represents the prior reset
request, not the in-flight run. Poll past it.scripts/setup-knowledge-base.sh (kept). Day-to-day
ops are the inline az/curl commands below.| Var | Used for |
|---|---|
AZURE_RESOURCE_GROUP | Resolves search service / storage |
AZURE_SEARCH_ENDPOINT | Override; otherwise https://<svc>.search.windows.net |
AZURE_SEARCH_ADMIN_KEY | Override; otherwise fetched via az search admin-key show |
STORAGE_ACCOUNT | Override; otherwise from azd outputs |
Get an admin key:
SVC=fibey-apps-search # or your env's search service
RG=rg-fibey-westus2
KEY=$(az search admin-key show --service-name "$SVC" --resource-group "$RG" --query primaryKey -o tsv)
These replace the previous wrapper scripts. Run from repo root with .env
exported (set -a && . .env && set +a).
SVC=fibey-apps-search
INDEX=foundry-iq-docs-index
INDEXER=foundry-iq-docs-indexer
EP="https://$SVC.search.windows.net"
KEY=$(az search admin-key show --service-name "$SVC" -g "$AZURE_RESOURCE_GROUP" --query primaryKey -o tsv)
# Doc count
curl -fsS -H "api-key: $KEY" \
"$EP/indexes/$INDEX/docs/\$count?api-version=2024-07-01"
# Latest indexer run
curl -fsS -H "api-key: $KEY" \
"$EP/indexers/$INDEXER/status?api-version=2024-07-01" \
| jq '.lastResult | {status, itemsProcessed, itemsFailed, errors, startTime, endTime}'
# Schema
curl -fsS -H "api-key: $KEY" \
"$EP/indexes/$INDEX?api-version=2024-07-01" \
| jq '{fields: [.fields[] | {name, type, key, retrievable, searchable}], semantic, vectorSearch}'
# Plain search (sanity check)
curl -fsS -H "api-key: $KEY" \
"$EP/indexes/$INDEX/docs?api-version=2024-07-01&search=*&\$top=3" | jq
ACCT=<your storage account>
CONT=foundry-iq-docs
KEY=$(az storage account keys list -g "$AZURE_RESOURCE_GROUP" -n "$ACCT" --query "[0].value" -o tsv)
# Upload (overwrites)
az storage blob upload-batch --account-name "$ACCT" --account-key "$KEY" \
--destination "$CONT" --source services/foundry-iq-docs/docs --pattern "*.md" --overwrite
# Run the indexer
SVC_KEY=$(az search admin-key show --service-name "$SVC" -g "$AZURE_RESOURCE_GROUP" --query primaryKey -o tsv)
curl -fsS -X POST -H "api-key: $SVC_KEY" \
"$EP/indexers/$INDEXER/run?api-version=2024-07-01"
# Poll until terminal (success / transientFailure / persistentFailure).
# IMPORTANT: ignore status "reset" (not terminal) and empty bodies.
for i in {1..60}; do
STATUS=$(curl -fsS -H "api-key: $SVC_KEY" \
"$EP/indexers/$INDEXER/status?api-version=2024-07-01" \
| jq -r '.lastResult.status // "running"')
echo "[$i] $STATUS"
case "$STATUS" in
success|transientFailure|persistentFailure) break ;;
esac
sleep 5
done
For a full rebuild, call POST .../indexers/$INDEXER/reset first, then run
the indexer. Don't bail on the first reset status while polling.
# KB config
curl -fsS -H "api-key: $SVC_KEY" \
"$EP/knowledgebases('fibey-field-ops-kb')?api-version=2026-04-01" | jq
# Retrieve
curl -fsS -X POST -H "api-key: $SVC_KEY" -H "Content-Type: application/json" \
"$EP/knowledgebases('fibey-field-ops-kb')/retrieve?api-version=2026-04-01" \
-d '{
"intents": [{ "search": "how do I splice single-mode fiber", "type": "semantic" }],
"knowledgeSourceParams": [{
"kind": "searchIndex",
"knowledgeSourceName": "fibey-field-ops-ks"
}]
}' | jq
lastResult.status == success and doc count matches expectations.success? Schema as expected?sourceDataFields.api-version=2024-07-01). Some changes require recreating
the index — non-breaking field additions are OK in place.scripts/setup-knowledge-base.sh so future bootstraps match.sourceDataFields or semantic config changed, also update the KS body
inside setup-knowledge-base.sh (and PUT to the KS endpoint).lastResult.status == persistentFailure repeatedly with the same error.search=* results.Recovery: POST /indexers/$INDEXER/reset, then re-run. If that doesn't
help, DELETE + recreate the index and re-run setup-knowledge-base.sh.
az search admin-key show./knowledgebases/<name>) — they return 405.
Use OData function syntax: knowledgebases('<name>').reset as terminal in your polling loop..env.example.