trajectory-review — Claude Code skill: what it does, install
trajectory-review is a Claude Code skill that examines an agent’s path to a result. It runs after a task or eval finishes, fails, or changes outcome.
MIT ★ 741 updated 2026-11-11
/plugin marketplace add MiniMax-AI/mcode-stack
/plugin install ronny@mcode-stackWhat trajectory-review does
trajectory-review reconstructs an agent run from a session transcript, tool-call log, or eval trace. It records the intent, key arguments, returned data, and next decision for each step. The review flags unchecked assumptions, detours, wrong tools or arguments, wasted context, ignored warnings, and broken earlier constraints.
It identifies the earliest step that made a failed run unlikely to succeed, or the riskiest step in a passing run. Each finding is tied to a concrete prompt, harness, or project-rule change, while isolated quirks stay out of shared rules. The run record is required for the review. Fixes are checked with harness-experiment by rerunning the task several times and comparing trajectories as well as outcomes.
What it helps with
- reconstruct an agent’s step-by-step execution from a run record
- locate the first decision that led a failed task off course
- separate checked evidence from assumptions made during execution
- spot repeated tool calls, idle polling, and other wasted turns
- map a trajectory issue to a prompt, harness, or project-rule fix
Example prompts for the agent
- Review the latest run from its transcript and mark every unchecked assumption.
- Build a compact timeline with each goal, tool call, response, and next decision.
- Find the riskiest step in this passing eval and suggest a check with
harness-experiment.
How to install trajectory-review
1Plugin marketplace
/plugin marketplace add MiniMax-AI/mcode-stack/plugin install ronny@mcode-stackThis installs the whole ronny plugin with all its skills
2Skills CLI (Vercel)
npx skills add MiniMax-AI/mcode-stack --skill trajectory-review -a claude-code3Manually: put the files in place
~/.claude/skills/trajectory-review/SKILL.mdCopy the whole skill folder; for one project use .claude/skills/trajectory-review/
1Skills CLI (Vercel)
npx skills add MiniMax-AI/mcode-stack --skill trajectory-review -a codex2Manually: put the files in place
.agents/skills/trajectory-review/SKILL.mdFor all projects use ~/.agents/skills/
1Skills CLI (Vercel)
npx skills add MiniMax-AI/mcode-stack --skill trajectory-review -a cursor2Manually: put the files in place
.cursor/skills/trajectory-review/SKILL.mdCursor also reads .agents/skills/ and .claude/skills/
1GitHub CLI
gh skill install MiniMax-AI/mcode-stack stacks/ronny/skills/trajectory-review2Manually: put the files in place
.github/skills/trajectory-review/SKILL.md1Skills CLI (Vercel)
npx skills add MiniMax-AI/mcode-stack --skill trajectory-review -a gemini-cli2Platform CLI
gemini skills install https://github.com/MiniMax-AI/mcode-stack --path stacks/ronny/skills/trajectory-reviewManually: put the files in place
.devin/skills/trajectory-review/SKILL.mdThe legacy .windsurf/skills/ path still works
Claude.ai
Zip the folder and upload it in Settings → Capabilities → Skills
Step by step, with commands for each platform: How to install skills in Claude Code: five ways
Questions about trajectory-review
How do I install trajectory-review in Claude Code?
Add the marketplace with /plugin marketplace add MiniMax-AI/mcode-stack, then install the plugin with /plugin install ronny@mcode-stack.
What does trajectory-review do?
It examines the recorded path behind an agent result, including tool use, detours, assumptions, missed signals, and instruction drift. The findings are turned into specific changes to prompts, the harness, or project rules.
What does trajectory-review need to work?
It needs an actual run record such as a session transcript, tool-call log, or eval trace. The review also records the model, settings, and repository state; fixes are checked with harness-experiment.
Source text
SKILL.md · MIT
---
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:
| Finding | Typical fix |
| --- | --- |
| Assumption never checked | A verification step in a skill or project rules |
| Could not find or picked the wrong tool | Tool description, naming or grouping (see `tool-budget-audit`) |
| Lost an earlier constraint | Write it into a file the agent reloads (see `long-session-check`) |
| Polling or waiting idle | Background tasks with notification (see `long-command-check`) |
| Repeated per-turn overhead | Context injection review (see `context-tax-audit`) |
| Model limitation | Record 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.
© Ronny (MiniMax Code team) · MIT · MiniMax-AI/mcode-stack