assignment-defend
Skill0 starsChanged 3 months agoMITView on GitHub
Optional add-on to assignment-mode for when the human will have to explain or defend the work to someone — a grader who asks questions, an oral defense, a code walkthrough. Adds explanation depth and defense prep on top of assignment-mode's verification baseline.
In this repository8 other files it gives its agents.
- assignment-modeSkill
- codegraphSkill
- commit-disciplineSkill
- debug-modeSkill
- health-reviewSkill
- learning-guardSkill
- teacherSkill
- spikeSkill
The file
We split this skill into the rules it gives an agent
3 rules. About 0 are common, 3 are distinctive, and 0 are very specific to this project.
≈0
Common · estimate
Many files like this have a rule of this kind.
≈3
Distinctive · estimate
This author's own choices. Worth a look.
0
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- Distinctive: line 35
After the solution is verified (per assignment-mode), pause and offer the defense sheet before considering the task finished.
- Distinctive: line 36
Prefer plain language and analogies over restating the code in prose.
- Distinctive: line 37
If the human says "just explain part X," go deep on X specifically rather than re-explaining the whole solution.
≈ “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
Assignment Defend
When to attach this
Some graders only run the code and check the output. Others sit the human down and ask them to explain any part of it. Attach this skill when it's the second kind — a professor who questions the submission, an oral defense, a viva, a code walkthrough as part of grading.
This skill assumes assignment-mode is already active and adds to it. It does not replace assignment-mode's verification requirement — a well-explained solution that doesn't actually work is still a failure.
Responsibilities
Claude MUST
- Explain the reasoning behind every non-trivial decision well enough that the human could reproduce or defend it without Claude present — the why, not just the what.
- Prepare a short defense sheet alongside the solution: 3–6 questions a grader plausibly asks about this specific submission, each with a concise answer the human can give in their own words.
- Offer — don't force — to quiz the human on the material the assignment covers, especially if they'll be questioned live.
- Walk through the code section by section in plain language on request, at whatever depth the human asks for.
- Point out the parts of the solution most likely to draw questions (a non-obvious algorithm choice, an edge case handled a specific way, a deliberate simplification) and explain them proactively, not just on request.
Claude MUST NOT
- Hand over an explanation so dense or jargon-heavy that the human can't actually repeat it in their own words — match the explanation to what a human would plausibly say out loud.
- Skip explaining a part just because it's small — a grader can ask about any line.
Working Style
- After the solution is verified (per assignment-mode), pause and offer the defense sheet before considering the task finished.
- Prefer plain language and analogies over restating the code in prose.
- If the human says "just explain part X," go deep on X specifically rather than re-explaining the whole solution.
Collaboration
Builds on assignment-mode — always active together, never standalone.
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.