[integration][auto-dig] actor failure in run 33253273594 #323
Labels
No labels
bug
documentation
duplicate
enhancement
good first issue
help wanted
invalid
question
wontfix
No milestone
No project
No assignees
1 participant
Notifications
Due date
No due date set.
Dependencies
No dependencies set.
Reference
nsaspy/prolog-rlm#323
Loading…
Add table
Add a link
Reference in a new issue
No description provided.
Delete branch "%!s()"
Deleting a branch is permanent. Although the deleted branch may continue to exist for a short time before it actually gets removed, it CANNOT be undone in most cases. Continue?
Auto-Dig integration failure diagnostics
lost-rob0t/starintel-gpt-auto-dig22d70678c0aa26c15c9e24d06ccea0d10dbcf885a89175bf925df3afa358f8d05a370fe2a03e2117Sanitized failure diagnostics
Summary:
phase=schema kind=unsupported_schema_type message=runtime schema type cannot be projected to provider JSON SchemaStructured Prolog-RLM error
Sanitized traceback / stderr tail (last 160 lines)
Root cause confirmed in
prolog/rlm_native_tool.pl/prolog/rlm_mcp_tool.pl: MCP import deliberately normalizes missing/unsupported JSON Schema types to runtimetype:any, but the native direct-mode provider schema renderer rejectsanyasunsupported_schema_type. This means valid MCP schemas with unconstrained/union-ish fields can import successfully and then fail before the first provider call when direct mode projects tools. Correct wire behavior for runtimeanyshould be an unconstrained JSON Schema (e.g.{}while preserving supported annotations such asdescription), not a literal provider type namedany. Add a regression test with an object argument property oftype:any, then allow that property to render as an unconstrained schema. Auto-Dig run 33253273594 reproduced this with the full 14-tool Brave+Fetch inventory; no model call was reached.