agentleFS
Bellistri-NickBellistri-Nick/pm-skills/skills/take-home-reviewer/SKILL.md

take-home-reviewer

Skill0 starsChanged 28 days agoMITView on GitHub

Review and sharpen a PM take-home assignment or case interview submission before the user sends it. Evaluates the work through the lens of what hiring teams actually look for: clear problem framing, defensible prioritization, realistic execution thinking, and a point of view. Scores the submission, identifies the highest-risk gaps, and offers targeted rewrites. Use this skill whenever the user has completed a PM take-home and wants feedback before submitting. Also triggers for: "PM take-home", "case assignment", "product case", "interview assignment", "PM homework", "review my submission", "take-home exercise", "product strategy assignment", "PM case study interview", or when the user shares a completed assignment and asks if it's ready to submit. Invoke immediately — gather context as part of the skill flow.

The file

We split this skill into the rules it gives an agent

6 rules. About 0 are common, 6 are distinctive, and 0 are very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈6

    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
  1. Distinctive:

    Ask for the following in one prompt.

    line 34
  2. Distinctive:

    Before scoring, identify what the assignment is actually testing.

    line 51
  3. Distinctive:

    After the fixes, offer:

    line 125
  4. Distinctive:

    When rewriting, preserve the user's voice and ideas.

    line 131
  5. Distinctive:

    If the submission has a real problem, name it plainly.

    line 168
  6. Distinctive:

    If the submission is strong, say so specifically — and still find the one thing that would make it stronger.

    line 170

≈ “Common” is an estimate: at least 1 in 5 of the 364,980 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.

Keep what fits and drop the rest. Your version, ready for your agents.
Files
SKILL.md

PM Take-Home Assignment Reviewer

What this skill does

Reviews a completed PM take-home or case assignment and gives the user honest, specific feedback before they submit. The goal is to catch the gaps that hiring teams will flag and surface the opportunities to show stronger PM thinking — without making the work sound like someone else wrote it.

Take-homes are high-stakes and time-pressured. This review simulates what a hiring panel actually looks for, calls out risks plainly, and offers targeted improvements the user can execute quickly.


Step 1: Gather context

Ask for the following in one prompt. If anything is already in context, skip asking for it:

  1. The assignment — Paste or describe the prompt/question
  2. The submission — Paste the full response or describe what they wrote
  3. The company and role — Who is this for? What stage (Series A, enterprise, B2B SaaS, consumer)? What role level?
  4. Time remaining — How long before submission? This affects how comprehensive the feedback can be
  5. Specific concerns — Anything they already suspect is weak

Once you have this, proceed. Don't wait for more context.


Step 2: Understand the assignment intent

Before scoring, identify what the assignment is actually testing. Most PM take-homes fall into one of these types — name which one (or which combination):

  • Product sense / prioritization — Given a product or metric, what would you do? Why? (Tests judgment, frameworks, user empathy)
  • Strategy / go-to-market — How would you enter this space / launch this product? (Tests market thinking, business model understanding)
  • Analytical / metrics — What's happening with this metric? What would you do? (Tests structured thinking, comfort with ambiguity)
  • Execution / roadmap — What would you build in the next 6 months / quarter? (Tests prioritization, trade-off reasoning, delivery thinking)
  • Product critique — What's wrong with this product? How would you improve it? (Tests user empathy, product intuition, communication)
  • 0-to-1 — Design a product for this user problem / market. (Tests vision, scoping, feasibility thinking)

Knowing the type shapes the scoring criteria. A prioritization exercise is graded differently than a metrics analysis.


Step 3: Score the submission

Score across five dimensions, each 1–10:

DimensionScoreWhat you're evaluating
Problem framingX/10Did they define the right problem? Is the scope clear?
Prioritization logicX/10Are the choices defensible? Is the reasoning explicit?
User and market understandingX/10Is there genuine insight about users/customers, or just assumptions?
Execution realismX/10Is the plan achievable? Are trade-offs acknowledged?
Communication clarityX/10Is it easy to follow? Does it show a clear POV?
OverallX/10Holistic signal: would a panel fight for this candidate?

For each dimension, provide:

  • A one-line verdict
  • The specific evidence from the submission (quote or reference)
  • The biggest risk if left unchanged

Step 4: Hiring panel simulation

Simulate what two reviewers will likely say when they debrief:

Senior PM Reviewer

What they'll notice first. What question they'll raise in debrief. Whether they'd advocate for moving this candidate forward. Be direct — "This passes" or "This gets flagged" with the reason.

Hiring Manager

What signals they're looking for at this level. Whether the submission demonstrates the right kind of thinking for the role. The thing that would make them hesitate.


Step 5: Top fixes — ordered by impact and feasibility

Given the time remaining, prioritize what to address first. Format each fix as:

[Fix title] What's wrong: [specific issue with reference to the submission] What to do: [concrete, actionable change] Time to fix: [~5 min / ~15 min / ~30 min] Why it matters: [what this unlocks in the panel review]

Order by: (1) gap severity × (2) fix difficulty. A 10-minute fix that closes a major gap beats a 2-hour polish that improves a minor one.


Step 6: Offer targeted rewrites

After the fixes, offer:

"Want me to rewrite any specific sections? I can sharpen the problem framing, strengthen the prioritization rationale, tighten the executive summary, or rewrite a section that isn't landing. Just say which one."

When rewriting, preserve the user's voice and ideas. The goal is to make their thinking clearer and more defensible — not to replace it.


Common failure modes to flag

Watch for these patterns — they appear often and hiring panels are trained to spot them:

Framework theater — Using 2x2s, rice/ice, or Kano without the judgment that should drive them. "I used RICE to score these features" is not enough. What were the inputs? Why those weights? What was the hardest call?

Missing the user — A submission that's all strategy and no users. Where is the evidence that real people have this problem?

Scope creep without acknowledgment — The submission answers a different (usually bigger) question than the one asked, without explaining why.

No point of view — A list of options with no recommendation. "We could do A or B" without committing signals indecisiveness, not thoroughness.

Undefended assumptions — The entire strategy rests on an assumption that's never named or tested. ("Assuming users want X...") Name assumptions explicitly and note how you'd validate them.

Polished but shallow — Well-formatted, lots of headers and bullets, but the actual thinking doesn't hold up to one question of "why." Formatting is not the work.


Tone guidance

This is a high-stakes review, not a cheerleading session.

  • If the submission has a real problem, name it plainly. "A panel will flag this" is more useful than "there's room to tighten."
  • If the submission is strong, say so specifically — and still find the one thing that would make it stronger.
  • Be direct about time constraints. If there are 2 hours left and five things to fix, name the one that matters most.
  • Match the urgency of someone who has been on hiring panels, has seen great submissions lose on execution, and actually wants this person to get the job.

Download now (ZIP)

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
What did you use it for? e.g. Reviewing pull requests in a Rails app

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.