it said there was to many steps for it
in progress
S
Student a3d1f821-08f
Category:
Conversations & ChatType:
User submissionUser role:
studentP
Patrick Jean
Re-verification (2026-05-27, commit 00eb58bb, feedback-investigator local session)
Bug confirmed still present at HEAD. Fix PR #1924 remains open and unmerged. recursion_limits.py:27
— TOP_LEVEL_CHAT_RECURSION_LIMIT=25
vs 42 entries in STUDENT_ALLOWED_TOOLS
. Multi-tool chains exceed the budget, triggering GraphRecursionError. Status: awaiting merge of PR #1924.P
Patrick Jean
Re-verification (2026-05-27, commit 00eb58bb, feedback-investigator local session)
Bug confirmed still present at HEAD. Fix PR #1924 remains open and unmerged. recursion_limits.py:27
— TOP_LEVEL_CHAT_RECURSION_LIMIT=25
vs 42 entries in STUDENT_ALLOWED_TOOLS
.S
Sage Investigator Bot
Re-verification (2026-05-27, commit 00eb58bb, feedback-investigator local session)
Bug confirmed still present at HEAD. Fix PR #1924 remains open and unmerged.
- recursion_limits.py:27—TOP_LEVEL_CHAT_RECURSION_LIMIT = 25
- tool_policy.py:34-90—STUDENT_ALLOWED_TOOLShas 42 entries
- Legitimate multi-tool chains (e.g., discovery session requiring topic lookup → mastery check → guided session start) can exhaust the 25-step budget, triggering the "too many steps" error.
Status remains
in progress
— awaiting merge of PR #1924.P
Patrick Jean
Re-verification (2026-05-27, commit 3fdf806d, run 26514124793)
Bug confirmed still present at HEAD.
TOP_LEVEL_CHAT_RECURSION_LIMIT
is still 25 in recursion_limits.py:27
while the student agent has 42 tools. Fix in PR #1924 (raise to 50) has not been merged.Status remains: in progress (awaiting PR #1924 merge).
P
Patrick Jean
[Investigator Bot] Re-verification (2026-05-27, commit 451a4786): Bug confirmed still present at HEAD. Fix PR #1924 remains open. recursion_limits.py:27 — TOP_LEVEL_CHAT_RECURSION_LIMIT=25 vs 42 entries in STUDENT_ALLOWED_TOOLS. Root cause unchanged. Status remains in progress.
S
Sage Investigator Bot
Re-verification (2026-05-27, commit 451a4786, feedback-investigator local session)
Bug confirmed still present at HEAD (unchanged since last check). Fix PR #1924 remains open and unmerged. - recursion_limits.py:27
— TOP_LEVEL_CHAT_RECURSION_LIMIT
is 25 but STUDENT_ALLOWED_TOOLS
has 42 entries — legitimate multi-tool workflows exceed the limit and surface the "too many steps" error to users.S
Sage Investigator Bot
Re-verification (2026-05-27, commit 451a4786, feedback-investigator local session)
Bug confirmed still present at HEAD (unchanged since last check). Fix PR #1924 remains open and unmerged.
- recursion_limits.py:27:TOP_LEVEL_CHAT_RECURSION_LIMIT = 25
- tool_policy.py:STUDENT_ALLOWED_TOOLShas 42 entries
- Legitimate multi-tool student conversations (discovery sessions, guided learning) can invoke many tools and exceed the 25-step limit, surfacing the "too many steps" error
S
Sage Investigator Bot
Re-verification (2026-05-27, commit 451a4786, feedback-investigator local session)
Bug confirmed still present at HEAD (unchanged since last check). Fix PR #1924 remains open and unmerged. recursion_limits.py:27
— TOP_LEVEL_CHAT_RECURSION_LIMIT
still 25 but STUDENT_ALLOWED_TOOLS
has 42 entries. Legitimate multi-tool conversations exceed the limit. PR comment added: https://github.com/momentiq-ai/sage3c/pull/1924#issuecomment-4550400508S
Sage Investigator Bot
Re-verification (2026-05-26, commit 451a4786, feedback-investigator local session)
Bug confirmed still present at HEAD. Fix PR #1924 remains open and unmerged. recursion_limits.py:27
— TOP_LEVEL_CHAT_RECURSION_LIMIT
is still 25 while STUDENT_ALLOWED_TOOLS
has 42 entries. Legitimate multi-tool conversations hit the limit and show "too many steps" error.P
Patrick Jean
[Investigator Bot] Re-verification (2026-05-26): Bug confirmed still present at commit 7316311. TOP_LEVEL_CHAT_RECURSION_LIMIT is 25 but STUDENT_ALLOWED_TOOLS has 42 entries — legitimate multi-tool chains exceed the limit and trigger GraphRecursionError. The error is caught with a friendly message, but the conversation still fails. Fix PR #1924 is open and addresses this by bumping the limit to 50 and adding a guard test. No new PR needed.
Load More
→