hashgraph-online/awesome-codex-plugins / plugins/alpoxdev/hypercore/skills/hyper-excel
hyper-excel
VerifiedUse this skill when the user asks to read, analyze, create, or modify Excel workbooks (.xlsx, .xlsm) or turn CSV data into spreadsheets - inspecting sheets and formulas, editing cells while keeping existing formatting and formulas intact, building formatted reports with charts, or auditing formulas for broken ranges and error values. Do not use for plain dataframes, PDF or slide output, or VBA macro authoring.
rules/verification.md
View on GitHub# Verification
> A workbook is not finished until the thing you claim about it was checked against the file.
## Gates, in order
| Gate | Command | Passes when | If it fails |
|---|---|---|---|
| Structural | `inspect <file>` | parses, no `E_*` warning | Report the failure; do not answer from the file |
| Coverage | `verify <file> [--max-cells N]` | `cells_complete == pass` | The sheet was cut short; every count covers only the cells read, so say so |
| Integrity | `verify <file>` | no `errorLiterals`, no mismatched cache, no cycle | List the cells; fix or disclose |
| Preservation | `patch …` output | `verification.status == "pass"` | Nothing was written; fix the operation |
| Read-back | re-`inspect` or an independent reader | opens, same sheet names, same cell values | Treat as a failed write |
| Cache | `recalc` then `verify` | `cache_complete == pass` | Report which formulas stayed uncached and why |
## What each gate actually proves
- **Structural** proves the package is readable and the parts the reader needs exist. It does not prove values are right.
- **Integrity** proves each recomputed formula agrees with the stored value, that no cell holds an error literal, and that no formula depends on itself. A formula the evaluator cannot compute is `unevaluated`, never `pass`, and one cell that makes the evaluator raise does not suppress the report for the others: `unevaluated` names the cell, the formula, and the error class.
- **Preservation** proves untouched parts are byte-identical. It does not prove the changed part is what the user wanted; the report does that.
- **Read-back** proves a second parser accepts the file. Without an independent reader installed, say the check was limited.
- **Cache** proves a reader that does not calculate sees a value. It does not make the cached value authoritative; `fullCalcOnLoad` keeps Excel recomputing from source.
## Numbers in the report
Three labels, used consistently:
- `캐시 값` — read from `<v>` in the file. State the cell.
- `재계산 값` — produced by the bundled evaluator. State the formula.
- `미검증` — the evaluator does not implement the function. State the function name.
A derived number (a growth rate, a difference) shows its inputs and formula, not just the result.
## What not to claim
- Do not say "수식 검증 완료" when some formulas were unevaluated.
- Do not say "서식 유지" from a visual glance; cite the digest comparison.
- Do not say "엑셀에서 열린다" unless a real spreadsheet engine opened it on this machine.
- Do not say "값이 맞다" when the only evidence is that the formula ran without error.
## Repair loop
Verification failure gets one repair and one re-verify. If the same check fails twice for the same reason, stop and report: what failed, what was tried, what the user must decide. Never retry blindly and never weaken a check to make it pass.
## Sources
> Checked 2026-11-13.
| Claim | Source |
|---|---|
| Gate order (structural, integrity, preservation, read-back, cache) | this package's design; each gate is one command in `assets/tools/` |
| `cache_complete` returns `partial` while any formula lacks a value | `verify_formulas.py`, read on 2026-11-13 |
| `fullCalcOnLoad` makes Excel recompute from source on open | <https://learn.microsoft.com/en-us/dotnet/api/documentformat.openxml.spreadsheet.calculationproperties.fullcalconload> |
| An independent read-back is available via `uv run --with openpyxl` | run on this machine 2026-11-13; openpyxl 3.1.5 |
### Evidence grade
`LOCAL` for the gate design and the tool fields; the recalculation row is Microsoft's documentation.
Related skills
| Skill | Stars | Updated | Works with | Status | Install |
|---|---|---|---|---|---|
| rea-analysis-optimizationmorluto/reaOptimize REA providers and shared analysis workflows for agent context, latency, CPU, and memory while preserving correctness and Evidence. Use for analysis bottlenecks and cache or query design; CI/test-suite optimization belongs to rea-ci-tests-audit. | Stars: 468k | Updated | Claude CodeCodex | Verified | Install |
| rea-ci-tests-auditmorluto/reaAudit REA CI and test suites for correctness, fixture drift, false passes, and measured performance; use for suite or workflow investigations and efficiency changes, rather than ordinary feature testing or a single test failure. | Stars: 468k | Updated | Claude CodeCodex | Verified | Install |
| rea-tool-designmorluto/reaAudit, investigate, design, or change REA investigation tools, CLI/MCP contracts, provider capabilities, and Evidence semantics. | Stars: 468k | Updated | Claude CodeCodex | Verified | Install |
| ponytail-auditDietrichGebert/ponytailQuality audit of the whole repo: bugs, security, real load, missing tests, speed, and what to delete. Most important first. | Stars: 194k | Updated | Claude CodeCodex | Verified | Install |