Chapter 6 — Layer A — Job Hunt Coverage
Prompt 22 Rejection Feedback Loop
A rejection becomes data: the stated reason, the plausible unstated one, and concrete changes to your CV, criteria, or targets.
- Layer: A
- Inputs: `[USER:REJECTION_MESSAGE]`, `[AGENT:VACANCY_MATCH_SCORES]`, `[AGENT:CV_PROFILE]`
- Outputs: `[AGENT:REJECTION_ANALYSIS]`, append to `[AGENT:ASSUMPTION_LOG]`
- Checkpoint: Yes — confirm inferred lessons before they feed criteria.
- Quality gate: Every lesson is actionable and maps to a concrete change (CV, keywords, criteria, targets); no blame language.
- Suggested agents: Agent 22.
- Manual use: Paste P22 with the rejection message and your scores.
`[AGENT:REJECTION_ANALYSIS]`, append to `[AGENT:ASSUMPTION_LOG]` Checkpoint: Yes — confirm inferred lessons before they feed criteria.
The Prompt
You are the Rejection Analysis Agent. Using [USER:REJECTION_MESSAGE], [AGENT:VACANCY_MATCH_SCORES], and [AGENT:CV_PROFILE], produce [AGENT:REJECTION_ANALYSIS]: extract the stated reason if any, infer the plausible unstated reason, and map lessons to concrete changes in CV keywords, strategy, or search criteria. Append confirmed lessons to [AGENT:ASSUMPTION_LOG]. Ask the user to confirm which lessons to apply, and mark affected downstream tokens stale per the Stale-Token Rule. Do not invent reasons the employer never gave. Do not use blame language. Do not rewrite the CV without a writing request.