grade-session
Skill1 starsChanged 5 months agoMITView on GitHub
Grade the session that's ending — two candid report cards (the user's prompting, the agent's performance) plus a coach note for each side. Prints in chat; optionally saves to a file.
In this repository33 other files it gives its agents.
- architecture-blueprintSkill
- burnSkill
- calm-ui-auditSkill
- capitalizeSkill
- cardSkill
- commit-mineSkill
- continuity-contextSkill
- elonizeSkill
- frontierSkill
- full-throttleSkill
- groundSkill
- harmonizeSkill
- harvestSkill
- honeSkill
- milkSkill
- mirrorSkill
- moonwalkSkill
- nightshiftSkill
- page-auditSkill
- pmfSkill
- postSkill
- promptSkill
- publishSkill
- razeSkill
- redlineSkill
- relaySkill
- snipSkill
- stashSkill
- tasteSkill
- tidySkill
- untilSkill
- vacationSkill
- vaultSkill
The file
We split this skill into the rules it gives an agent
1 rule. About 0 are common, 0 are distinctive, and 1 is very specific to this project.
≈0
Common · estimate
Many files like this have a rule of this kind.
≈0
Distinctive · estimate
This author's own choices. Worth a look.
1
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- Very specific: line 32
If $ARGUMENTS gives a path, also save the full report there.
≈ “Common” is an estimate: at least 1 in 5 of the 350,495 skill files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.
Files
You are grading the session that is ending, from what you lived — do NOT re-read the transcript.
The grades
Two tables. Each row: Dimension | Grade (A–F, +/− allowed) | Notes (one candid sentence, specific to THIS session).
User Communication
- Prompt Clarity — could you act on first read, or did you have to guess?
- Context Provided — did they give what you needed, or did you have to dig?
- Specificity of Intent — did asks state the goal, or just an artifact?
- Feedback Quality — when correcting you, did they say what was wrong and why?
Agent Performance (grading yourself — be harder here than on the user)
- Output Quality — correct, complete, idiomatic?
- Alignment to Intent — built what was meant, not just what was said?
- Depth — surfaced non-obvious considerations, or stayed shallow?
- Efficiency — wasted turns, redundant work, questions you could have answered yourself?
- Avoided Mistakes — count them honestly, including ones the user never noticed.
The notes
- Coach Note (for user) — 2–3 sentences: the single highest-leverage change to how they prompt and steer, grounded in a real moment from this session.
- Agent Note — what you'd do differently next session; name your own errors plainly.
Rules
- Grade candidly. A report card of straight A's is a failed grading.
- Evidence over vibes: every grade must be justifiable by a specific moment from the session.
- If
$ARGUMENTSgives a path, also save the full report there; otherwise just print it.
Does it work?
Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.
No reviews yet
People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.
- It worked
- Partly
- It didn't
Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.
Your trail
Files you open will collect here, so you can find them again.
Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.
Now do this for your own.
Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.