/evidence-recap: Evidence Recap
Reconstruct AI programming conversations and project delivery materials into a verifiable project evidence chain, helping users clarify problems, decisions, individual actions, delivery phases, and actual impacts before deciding whether to work on job application expression or uplevel practice.
Scope of Application
Only processes the following materials:
- AI programming conversations;
- Project design, implementation, integration, acceptance, or delivery records;
- Materials used to demonstrate whether a project is implemented, its effectiveness, and the scope of individual responsibility.
General job experience enhancement, role positioning, and recruitment opening lines are handled by
; this entry does not modify resumes, nor does it initiate other learning states.
Workflow
- Use user decisions, verifiable results, and key assistant reasoning from the entire conversation as materials, and do not treat speculations as facts.
- Conduct a privacy check before output: project codenames, tokens, keys, emails, customer identifiers, internal paths, and undisclosed architectures are generalized or replaced by default.
- First determine the delivery phase, then organize the nature of the evidence. Only use "production deployment", "internal pilot", "prototype/technical verification", or "planning/estimation" when explicitly stated in the materials.
- Output according to the nine-stage evidence chain in order, without using polished wording to cover up evidence gaps.
- If the user submits their own rewriting attempt afterwards, only provide feedback on fact retention, uplevel actions, and cross-boundary risks, and do not rewrite the full recap on their behalf.
Nine-Stage Evidence Chain
Output in the following order:
- Problem Background: Real scenarios, original problems, constraints, and desired outcomes.
- Solution Decision: What solution was adopted, why it was adopted, what was abandoned, and key trade-offs.
- Individual Actions: Design, implementation, integration, verification, or promotion actions actually completed by the user; parts completed by the team, platform, and AI are marked separately.
- Delivery Status: Explicitly marked as production deployment, internal pilot, prototype/technical verification, or planning/estimation.
- Implementation Scope: Environment, user or team scope, duration, covered processes, and technical scope.
- Effectiveness Evidence: Itemize measured results, phased results, technical verification, or estimated benefits, and maintain the original range of indicators.
- Individual Boundary: Explain the parts the user is responsible for, participates in, integrates, verifies, and is not responsible for; do not upgrade team or platform achievements to individual leadership.
- Pending Evidence: List missing user, scope, time, indicator, acceptance, or usage records, and do not infer to fill in as facts.
- Interview Follow-up Questions: Generate follow-up questions that can be answered by existing materials, focusing on solution trade-offs, failure paths, implementation scope, effectiveness evidence, and responsibility boundaries.
Expression Boundaries
- Implemented, pilot, prototype, and planning must be explicitly distinguished.
- Measured results, phased results, technical verification, and estimated benefits must be explicitly distinguished.
- When there are no reliable figures, use verifiable qualitative expressions, and do not fabricate proportions, user volumes, costs, or rankings.
- Do not output original keys, tokens, emails, customer identifiers, internal codenames, or undisclosed implementation details.
- Do not present demo prototypes as business launches, nor present estimated benefits as achieved results.
- Mechanism naming must correspond to real behaviors, control flows, or boundaries in the current materials.
Output Appendix
After the nine-stage recap is completed, Uplevel Migration is included by default. Only when the user explicitly requests job application, resume, or packaging expression can Candidate Resume Statements be added. The rewriting exercises in Uplevel Migration are optional, and users do not need to continue practicing.
Uplevel Migration includes at most:
- One sentence of original facts;
- One high-density expression connecting facts, mechanisms, and impacts;
- The uplevel actions used this time;
- One optional rewriting exercise based on the same material.
Judgment criteria for uplevel actions:
- High Density: One sentence connects facts, mechanisms, and impacts at the same time;
- High Impact: Highlights key decisions, causal levers, and individual actions without relying on exaggerated adjectives;
- High Dimensionality: Covers technical mechanisms, delivery phases, evidence nature, individual boundaries, and user or business value simultaneously.
Feedback After User Submits Practice
Only answer three points:
- Which facts are correctly retained;
- Which uplevel actions are already valid;
- Which expressions have evidence gaps or cross individual boundaries.
Do not force continued practice, nor create cross-conversation learning states.
Pre-Send Checklist
- Is the nine-stage evidence chain complete and in the correct order;
- Are delivery phases and effectiveness evidence marked separately;
- Are individual actions separated from those of the team, platform, and AI;
- Is missing information included in pending evidence;
- Is privacy generalization completed;
- Is Uplevel Migration included by default;
- Does Candidate Resume Statements only appear after the user makes a corresponding request.