跳至正文
OpenAgentSkill
浏览已开放 · 目录记录日期 2026-11-21 · 自动更新和账户服务暂时关闭。
其他技能

trajectory-review

Review how an agent reached its result, not only whether the result passed, covering tools called, detours, unverified assumptions and wasted round trips, and turn the findings into concrete harness or prompt fixes. Use after an agent task finishes or fails, when an eval task flips, or when asked to look at the process behind a result.

MiniMax-AI · 已有静态检查记录

安装与使用

运行命令前先阅读来源与依赖说明。复制命令不会自动安装或执行。

npx skills add MiniMax-AI/mcode-stack --skill trajectory-review

技能说明


name: trajectory-review description: Review how an agent reached its result, not only whether the result passed, covering tools called, detours, unverified assumptions and wasted round trips, and turn the findings into concrete harness or prompt fixes. Use after an agent task finishes or fails, when an eval task flips, or when asked to look at the process behind a result.

Trajectory Review

A passing result can hide a fragile process: an assumption that happened to hold, a lucky retry, ten round trips where two would do. A failing result usually has its cause several steps before the error. Review the path, not only the end point.

1. Get the real trajectory

Use the actual record of the run: the session transcript, tool-call log, or eval trace. Note the model, settings and repository state. Do not reconstruct the run from memory.

2. Walk it step by step

Build a compact timeline. For each step: what the agent intended, which tool it called with which key arguments, what came back, and what it decided next.

While walking, mark:

  • Unverified assumptions. The agent states or acts on something it never checked ("callers already validate this", "the test passes").
  • Detours. Loops, repeated searches with changing keywords, re-reading the same file, retries without a change in approach.
  • Wrong tool or wrong arguments. Including tools it could not find.
  • Context waste. Large outputs pulled in and never used, polling instead of waiting for a notification.
  • Missed signals. Errors, warnings or test failures it saw and ignored.
  • Instruction drift. Earlier constraints it later broke, often after compaction.

3. Find the first point where it went wrong

For a failure, find the earliest step after which success became unlikely. For a pass, find the riskiest step: where a slightly different input would have broken it.

4. Turn findings into fixes

Map each finding to the cheapest durable fix:

FindingTypical fix
Assumption never checkedA verification step in a skill or project rules
Could not find or picked the wrong toolTool description, naming or grouping (see tool-budget-audit)
Lost an earlier constraintWrite it into a file the agent reloads (see long-session-check)
Polling or waiting idleBackground tasks with notification (see long-command-check)
Repeated per-turn overheadContext injection review (see context-tax-audit)
Model limitationRecord it; re-check after the next model upgrade

Keep one-off quirks out of project-wide rules. Only propose a rule for something that has happened more than once or caused real damage.

5. Confirm

A fix is a harness change, so check it with harness-experiment: rerun the same task several times and compare the trajectory, not only the outcome.

Output

The compact timeline with marked steps, the first wrong (or riskiest) step, and a short list of proposed fixes, each tied to a specific step.