eligibility-check

name: eligibility-check description: Use after /parse-rfp to determine whether Consultadd qualifies to bid this RFP, what gaps exist, and whether gaps can be closed via partnership / license rental / sub-vendor / teaming agreement. CHALLENGE GATE — hard-blocks the orchestrator on human approval if any gap exists. Triggers on "check eligibility", "can we bid this", "qualification check".

Eligibility Check — CHALLENGE GATE

Determine whether Consultadd is qualified to respond to this RFP. Compares hard requirements (NAICS, set-asides, certifications, clearances, past-performance thresholds, insurance, bonding, geography) against knowledge/certifications.md + knowledge/capabilities.md + knowledge/past-wins/*. Outputs an explicit qualified / mitigatable / no-bid recommendation. Hard-blocks the orchestrator until the analyst makes an explicit decision on every gap.

Workflow

1. Read sources.

   • ./rfp-parsed.md (RFP) and ./rfp-meta.json (metadata)
   • ${CLAUDE_PLUGIN_ROOT}/skills/_shared/knowledge/certifications.md
   • ${CLAUDE_PLUGIN_ROOT}/skills/_shared/knowledge/capabilities.md
   • ${CLAUDE_PLUGIN_ROOT}/skills/_shared/knowledge/past-wins/*.md (for past-performance threshold matching)

2. Extract hard requirements from the RFP. Build a structured list:

   • NAICS code(s) required
   • Set-aside status required (e.g., 8(a), HUBZone, WOSB, SDVOSB) or "full and open"
   • Certifications required (ISO, CMMI, SOC 2, etc.)
   • Security clearance required (facility level, personnel levels + counts)
   • Past performance threshold (often "$X in similar contracts in last Y years" or "Z prior engagements of similar scope")
   • Insurance minimums (general liability $X, professional liability $Y, cyber $Z)
   • Bonding (bid bond, performance bond, payment bond — amounts)
   • Geography (registered to do business in state X, local-presence requirement)
   • Workforce composition (% local hire, DBE/MBE/WBE participation)
   • Software / platform certifications (specific vendor partnerships, FedRAMP, StateRAMP)
   • Other (anything else stated as mandatory — "shall", "must", "required")

3. Compare each requirement against knowledge/. For each, mark:

   • MET — knowledge file confirms we satisfy
   • GAP — we don't satisfy, plus assessment of mitigation:
     • partnership — large-business partner can prime, we sub
     • sub-vendor — we prime, sub the gap-filling capability
     • teaming agreement — joint venture for this specific bid
     • license rental — buy/rent the missing certification or vendor partnership
     • not mitigatable — gap is structural; recommend no-bid
   • UNCLEAR — knowledge file insufficient to determine; needs human research

4. Past-performance threshold check. If the RFP requires N prior engagements of $X+ in within , count matching knowledge/past-wins/*.md files. Honor Approved for Use In filters.

5. Output ./eligibility-report.md. Structure:

   # Eligibility Report — <RFP Number>
   
   **Recommendation:** QUALIFIED | MITIGATABLE | NO-BID
   
   ## Met Requirements
   - <list, each with knowledge-source citation>
   
   ## Gaps (Mitigatable)
   ### <Requirement Name>
   - **What's required:** <RFP quote>
   - **Our status:** <gap detail>
   - **Suggested mitigation:** <partnership | sub-vendor | teaming | license rental>
   - **Recommended partner candidates:** <if any in past-wins or known network>
   - **Cost / effort to close:** <rough estimate>
   
   ## Gaps (Not Mitigatable)
   - <list — these would force a no-bid recommendation>
   
   ## Unclear Requirements (Need Human Research)
   - <list>
   

6. CHALLENGE GATE — hard block. Output a ./eligibility-decision-required.md with the explicit decision the analyst must make:

   # DECISION REQUIRED — Eligibility
   
   The orchestrator is paused. Before /rfp continues, you must explicitly resolve each gap below.
   
   ## Gaps Awaiting Decision
   1. <Gap name> — choose one: [confirm mitigation: <option>] [accept risk] [no-bid]
   2. ...
   
   ## Recommendation
   <model's recommendation, but the decision is yours>
   
   ## How to Resume
   After deciding, run `/rfp --resume` (or invoke the next phase manually).
   

   The orchestrator MUST NOT proceed past this gate until the analyst writes their decision. Implementation: orchestrator checks for ./eligibility-decision.md (analyst-authored) before continuing — if absent, stop.

7. Emit telemetry.

   (Telemetry is out of scope for v1. When the telemetry MCP Connector is live in v1.5, this skill will call its emit tool here.)

Constraints

• Hard requirements only. Don't list nice-to-haves or scoring criteria — those are for sales-mktg-expert and the response itself.
• No fabricating mitigation paths. Suggested partners come from past-wins files or are flagged as "needs research" — never invented.
• Recommendation must be unambiguous. QUALIFIED / MITIGATABLE / NO-BID. No "probably qualified" hedging.
• Past-performance citation must honor Approved for Use In. Don't count an excluded past win toward the threshold.
• Hard block is non-negotiable. Never let the orchestrator skip past a real gap.

Output Files

• ./eligibility-report.md — full analysis
• ./eligibility-decision-required.md — created only if gaps exist; orchestrator blocks on this
• ./eligibility-decision.md — analyst-authored response (presence unblocks orchestrator)