Skip to content
skillmaster
FREnglish version : EN
Install

reasoning-trace-optimizer

muratcankoylan/Agent-Skills-for-Context-Engineering / examples/interleaved-thinking

analyzes agent traces , which improves: spots drift and improves its prompts

score 4/10high risk18.2K repo stars
View on GitHub (new tab)

Details

judge’s score
4/10
place
outside the packs
publisher
muratcankoylan
status
new
pinned commit
9ff0c99 on GitHub (new tab)
skill’s last commit
May 15, 2026
repo stars
18,177
main type
Building with Claude

Judgement

Vague triggering, focused on MiniMax M2.1 (API, interleaved thinking). The rto Python package and the analyze-session mode are neither provided nor implemented for Claude Code. Useful ideas, but little verifiable know-how.

reservations

  • The rto Python package (cli, analyzer, optimizer) is not provided: commands and actual behavior cannot be verified.
  • Written for the MiniMax API (api.minimax.io, M2.1), not for Claude Code. The analyze-session mode and the post_tool_error hook event match no documented Claude Code API.
  • Partly marketing content (Partnership section, quotes from the MiniMax article) and truncated annexes (agentthinking, interleavedthinking, analysis); the interleavedthinking annex remains largely unread.
sub-scores
Triggering
1/2
Added value
1/2
Precision
1/2
Scope
1/2
Quality
0/2

Alerts

3 alerts, by category

Execution

  • high risk

    The example hook runs a shell command (rto analyze-session) on a tool error: if copied as is, it executes third-party code on every error.

Network

  • high risk

    The optimization mode sends tasks, prompts and full traces (which may contain code or secrets) to a third party, the MiniMax API.

Static scan

  • info risk

    bundled scripts, not run by the SKILL.md

    examples/01_basic_capture.py, examples/02_tool_usage.py, examples/03_full_optimization.py, reasoning_trace_optimizer/__init__.py, reasoning_trace_optimizer/analyzer.py

Original description

Debug and optimize AI agents by analyzing reasoning traces, context degradation, tool confusion, instruction drift, repeated task failures, and performance regressions.

added on Oct 19, 2026 · license MIT