Souls: web research guidance — Firecrawl-backed web_search works

Verified end-to-end from a worker run: web toolset provider switched to
Firecrawl Self-Hosted (the GUI env var was set but the provider selection
was never applied; PUT /api/tools/toolsets/web/provider committed it) and
a single web_search(query=...) call returned live results. Souls now tell
workers web_search/web_extract work and require a non-empty query.
Redeployed to all live profiles.

Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
This commit is contained in:
fez5587andClaude Fable 5 committed 2026-08-04 17:38:30 -05:00
1 parent 8a062af60f
commit 128af5d7af
1 file changed
+5
+5
View File
@@ -31,6 +31,11 @@ result="Wrote /workspace/index.html (3 KB): headline, three feature cards,
email form. Re-read file; all tags closed. No blockers.") email form. Re-read file; all tags closed. No blockers.")
A run that ends without calling kanban_complete or kanban_block counts as A run that ends without calling kanban_complete or kanban_block counts as
failed, no matter what work was done. failed, no matter what work was done.
## Web research
web_search and web_extract work (Firecrawl-backed). Always pass a non-empty
query string to web_search — an empty query is rejected. Prefer them over
browser_* tools for reading public pages.
""" """
ROLE_CHARTERS = { ROLE_CHARTERS = {