rejection-review
Skill312 starsChanged 10 days agoMITView on GitHub
A structured, honest post-mortem after an interview or OA, whether rejected, ghosted, or unsure. It reconstructs what happened round by round, finds the real root causes (knowledge gap, problem-solving process, communication, time, nerves, or fit), cross-checks them against past cracked sessions for repeating patterns, and turns them into three targeted drills and an updated plan. Use when the user got rejected, failed an interview or OA, wants to debrief an interview, or asks "why do I keep failing interviews".
In this repository20 other files it gives its agents.
- AGENTS.mdAGENTS.md
- website/AGENTS.mdAGENTS.md
- company-intelSkill
- cracked-profileSkill
- setup-crackedSkill
- core-vivaSkill
- defend-my-projectSkill
- hr-roundSkill
- mock-interviewSkill
- oa-simSkill
- pattern-coachSkill
- system-design-mockSkill
- github-glowupSkill
- jd-matchSkill
- make-it-hireableSkill
- referral-dmSkill
- resume-blueprintSkill
- roast-my-resumeSkill
- offer-decodeSkill
- prep-planSkill
The file
We split this skill into the rules it gives an agent
3 rules. About 0 are common, 1 is distinctive, and 2 are very specific to this project.
≈0
Common · estimate
Many files like this have a rule of this kind.
≈1
Distinctive · estimate
This author's own choices. Worth a look.
2
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- Very specific: line 14
Call the Skill tool with "cracked-profile", then read $CRACKED_DIR/journal.md.
- Distinctive: line 18
Ask in one round (numbered.
- Very specific: line 59
If $CRACKED_DIR/plan.md exists, offer to run /prep-plan to fold the drills into the plan.
≈ “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.
Files
Rejection review
Rejection hurts, and most candidates respond by grinding more random problems. The fix is almost never "more of everything". It's one or two specific failure modes, repeating. Find them.
0. Open with care, briefly
One line of acknowledgement, no therapy speak: "That's frustrating. Let's make it count: a 10-minute debrief now beats 100 random problems."
Call the Skill tool with "cracked-profile", then read $CRACKED_DIR/journal.md.
1. Reconstruct, while it's fresh
Ask in one round (numbered; they can answer briefly):
- Company, role, and which round(s): OA, technical (how many), system design, HR/managerial.
- For each round: the questions asked, as close to verbatim as they remember.
- What they did: approach, how far they got, whether the code ran, how much time was left.
- Interviewer signals: hints given, follow-ups, tone, the moment it felt like it went wrong.
- Their own gut feeling: where do they think it went wrong?
Then ask follow-ups on the round that most likely decided it. Up to 3 questions, one at a time.
2. Diagnose
Classify the evidence against these failure modes. A rejection usually has one primary cause and maybe a secondary one:
| Failure mode | Typical evidence |
| Knowledge gap | Didn't know the pattern, concept, or theory at all |
| Problem-solving process | Knew the topic but jumped to code, no brute force first, no examples, got lost |
| Implementation | Right approach, but buggy code, edge cases missed, couldn't debug |
| Communication | Silent for long stretches, unclear explanations, didn't ask clarifying questions |
| Time management | Stuck too long on part 1, never reached the follow-up |
| Nerves / performance | Knew it at home, blanked in the room; this often shows as a big home-vs-interview gap |
| Fit / external | Role or level mismatch, a strong competing pool, headcount, or no clear performance failure |
Be honest. If the evidence says "you didn't know Dijkstra", say that. If it genuinely looks external, say that too, so they don't over-correct. Some rejections aren't about performance.
Cross-check the journal: normalise its tags onto the canonical list in cracked-profile, then check whether these failure modes or weak-spot tags appear in past mocks. A repeat is the headline finding: "This is the third time edge cases sank you: twice in mocks, now in a real round."
3. Prescribe
- Root cause, in 1–2 sentences.
- Three drills, each specific, time-boxed, and mapped to a cracked skill where possible. For example:
- "Edge cases: for the next 10 problems, write 5 edge-case inputs before coding. Then run
/mock-interview. Target: Correctness ≥ 8." - "Graph shortest paths:
/pattern-coach shortest path, then 6 problems from its ladder over 4 days." - "Nerves: three
/mock-interviewsessions this week, timed and out loud. Exposure is the fix."
- "Edge cases: for the next 10 problems, write 5 edge-case inputs before coding. Then run
- What went right: at least one real positive from the evidence (not filler).
- Re-attempt: give the key question(s) from the interview back as a practice problem now, or schedule it, so they close the loop.
4. Log and re-plan
- Log a scorecard via the Skill tool with "cracked-profile", titled
<Company> · <round>. Dimensions: Knowledge, Problem-Solving Process, Implementation, Communication, Composure, scored from the reconstruction. Tag weak spots precisely; this is whereprep-plangets its data. - If
$CRACKED_DIR/plan.mdexists, offer to run/prep-planto fold the drills into the plan. - Close with one line of perspective that's true and not saccharine, e.g. "Most people who get offers have a pile of rejections behind them. The difference is that they debriefed."
Skill calls: if your agent has no Skill tool, open the named skill's SKILL.md in the cracked skills folder (core/, interview/, profile/, strategy/) and follow it instead.
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.

