anthropics / anthropics/claude-code
Model issued MCP trace calls by bare symbol name without uids; five of six failed
- Dominant language
- Python
- Stars
- 145k
- Forks
- 23.1k
- PR merge metrics
- PR metrics pending
Description
**What happened**
The model called the GitNexus trace tool with common names (execute, run, start, execution) plus file hints. Results: one path found, three no_path, one ambiguous, one truncated. Earlier context results had returned exact uids the model did not reuse.
**Expected**
from_uid and to_uid from prior results, and a lower depth.
**Repro**
Call trace with names shared by many symbols in a large repository.
**Environment:** Claude Code CLI 2.1.270 (Claude Code), macOS, auto mode, model claude-fable-5-1, session of 2026-09-14.
Contributor guide
No contributing guide indexed for this repository
Research direction
No source file or test is named. Start by reproducing the MCP trace call with common symbol names and file hints in Claude Code, then compare it with a call using exact UIDs from earlier context results and a lower depth. Done means the model reliably reuses those UIDs and avoids the reported no_path, ambiguous, and truncated outcomes.
Written by the indexing model from the issue text.
Assessment
- Tech stack
- python
- Domain
- cli, devtools
- Issue type
- Bug
- Difficulty
- 4/5
- Estimated time
- 3-5 days
- Activity status
- Active
- Clarity
- Needs clarification
- Newbie friendliness
- 35/100