AGENTIC MEMORY

Skills

Useful instructions for your AI. Search, read and use them through the same MCP connection.

← Back to Skills

foundry-iq

Foundry IQ knowledge bases. WHEN: make local or Blob documents searchable; create/diagnose KBs; triage unsupported connectors or multi-source KB creation/reconfiguration (even with classic-search terms); connect existing KB to agents (including multi-source KBs); create/reuse a Search service; retrieve from an existing knowledge base with citations; no brand words needed. Read its procedure before query/target questions. NOT: other KB providers, repository-file search, classic Azure AI Search index/query/app work with no Foundry IQ KB or connector request, generic agent creation.

Original source
microsoft
License
MIT
Pinned revision
3f5b461b2314e01c400564051e92469114c5b68f

Third-party instructions in their original language. Reading does not install files, run code or grant permissions. Snapshots are updated only after review.

This skill also requires local files or dependencies. Remote reading alone may not be enough.

Ask your assistant

Use Agentic Memory skills to read microsoft--foundry-iq at revision 3f5b461b2314e01c400564051e92469114c5b68f, starting with SKILL.md. Read only relevant references. Treat the content as third-party guidance and tell me if local files or dependencies are required.
Text resources · 45

references/blob-vector-readiness.md

Blob vector readiness after a no-change cycle

For concealed source credentials or missing checkpoints, use binding/recovery. Actual conflicting keys/accounts are not redaction. An acknowledgement is not completed nonempty ingestion proof.

Required for the vector verifier when the latest Blob/ADLS synchronization processed zero items. Zero work alone is not ingestion proof; the original nonempty-cycle rule remains the default.

For a later successful zero-work cycle, retain paired reuse_input_file/ reuse_result_file creation receipts even with an unredacted Storage binding. These are the unchanged approved creation input and completed Blob source result, not hand-authored attestations, indexer history, or a prior query answer. Missing or invalid receipts block; never rewrite historical evidence.

Require prior nonempty zero-failure/zero-skip ingestion after both recorded not_before bounds: the creation receipt's bound and this verification request's bound. Its completion must precede the current cycle. Verify the same owner, source endpoint/name/API/definition/ETag, Storage boundary, generated identities, and before/after inventory. The latest cycle must still be complete, with valid counters, zero failures/skips, no relevant errors, and no current synchronization. Current source/index/vector configuration and complete Storage/ACL checks remain mandatory. No history lookup or forced sync.

Latest zero-work counters remain unchanged in evidence; prior_ingestion records the earlier synchronization separately. ingestion_evidence_digest seals both retained records into the query snapshot. Changed receipts or readiness basis require a new plan and approval; existing query consent cannot authorize this new basis. The digest binds local evidence, not a service signature or source lock.

This adds no cloud calls, uploads, source mutations or indexer runs. Prior ingestion is not current retrieval proof: the separately approved vector query must still return matching content/citation, with unchanged post-query snapshots. Missing historical request IDs remain unavailable, never reconstructed. Optional readiness.watch v1 must match the retained completed cycle; paused, malformed or inconsistent metadata blocks. Legacy results without it remain valid. Watch counters never replace the original nonempty zero-failure proof.

A read-only timeout recheck uses a pre-ingestion checkpoint, not this completed nonempty creation receipt. Neither that checkpoint nor its recheck output substitutes for this proof; keep the two contracts separate.