Generate XLSX

⚠ TEMPLATE ONLY — Slot for XLSX generation

This skill is a scaffolding slot per L-001 (Lintel = scaffolding, not curated content). The frontmatter + agent-mapping entry exist; the workflow body does not. When an operator invokes this skill, the AI generates the appropriate workflow content at invocation time using the design-spec.json + content.md from the shared pipeline.

Why a slot exists

XLSX is the natural artifact for tabular data: cost-estimates, capacity-forecasts, comparison-matrices, customer-question-trackers. The shared pipeline's outline → write produces text content; XLSX-generation requires identifying which sections are tabular candidates + structuring them as worksheets with formulas. This decision is content-specific and benefits from AI-at-invocation rather than pre-baked workflow.

At-invocation contract

When invoked (via /li:generate-xlsx --from-pipeline <run-dir> or directly):

1. Tabular extraction: AI reads content.md + identifies sections that are inherently tabular (estimates, comparisons, schedules, options-matrices). Non-tabular sections (narrative, vision) are excluded.
2. Worksheet structure: One worksheet per tabular section + one summary worksheet linking them. Use design-spec.json's palette for header colors + font for consistency with sibling artifacts.
3. Formula insertion: Where estimates appear, AI inserts SUM/PRODUCT/AVERAGE formulas instead of hardcoded totals. Where comparisons appear, AI uses conditional-format rules for red/green status.
4. Build: Use openpyxl or equivalent Node-based xlsx library to produce *.xlsx. Brand-template from ~/.lintel/brand/xlsx-templates/ if present.
5. QA hand-off: Pass produced *.xlsx to /li:generate-qa for validation.

Brand template

~/.lintel/brand/xlsx-templates/ slot exists. Templates might include: customer-estimate-template.xlsx, engagement-tracker-template.xlsx, comparison-matrix-template.xlsx. If template present, AI uses it as starting structure. If empty, AI creates fresh workbook from design-spec.

Agent dispatch

Per skills/generate/agent-mapping.yaml:

• Primary: CostAnalyzer (XLSX use cases skew cost/estimate)
• Conditional: CapacityPlanner if capacity/sizing content
• Conditional: SecurityAuditor if content references customer-sensitive data

Voice tier

voice: internal default. XLSX content typically internal-only; even when shared with customer, voice is data-driven (numbers + labels), not prose.

If invoked with --customer-share, requires an upstream PASS from the active pack's voice gate (none by default) on any prose labels/headers (orchestrator-level gate).

Status protocol

• DONE — *.xlsx produced + qa-handoff successful + formulas validated
• DONE_WITH_CONCERNS — produced but qa flagged formula errors or layout issues
• BLOCKED — content.md has no tabular sections + no fresh-data strategy declared
• NEEDS_CONTEXT — --from-pipeline directory missing or design-spec absent

When to promote from slot to curated

If a recurring estimate-pattern emerges (e.g., always the same cost-estimate with same worksheet structure), promote that pattern into the SKILL.md body — that becomes a canonical XLSX-builder for that scenario.

Until then: AI generates fresh per invocation. Repo stays clean.

Recommended next steps

After invocation:

• For QA: /li:generate-qa --artifacts <xlsx-path>
• For multi-format consistency: invoke parent /li:generate instead of solo-invocation
• For data-fresh: re-invoke with --update-data <source> to refresh estimates against current pricing/inventory