tabular-cleanup
gaasher/Agent-Loop-Skills / loops/tabular-cleanup
cleans a messy table against an inferred contract , which improves: fewer errors, deterministic and traceable checks
Details
- judge’s score
- 9/10
- place
- outside the packs
- publisher
- gaasher
- status
- new
- pinned commit
- f1169e6 on GitHub (new tab)
- skill’s last commit
- Jun 22, 2026
- repo stars
- 175
- main type
- Data
Judgement
Very clear triggering with explicit exclusions, precise cleaning loop (deterministic checks, safeguards, defined stop). Score reduced because the cited appendices (schemas, example) are missing and the text is very long.
reservations
- Cited appendices missing: examples/run.example.yaml, schemas/ledger.schema.json and schemas/profile.schema.json were not provided, so the formats cannot be verified.
- Long and dense text (contract, checklist, loop, stops, formats); reading it in full is costly in context.
- The skill does not specify the exact minimum dependencies or the pandas version tested.
- Triggering
- 2/2
- Added value
- 2/2
- Precision
- 2/2
- Scope
- 2/2
- Quality
- 1/2
Alerts
1 alert, by category
Execution
- high risk
The skill has agent-generated transformation code run in the user's environment via <analysis_cmd>, without human review after contract validation.
Original description
Use when the user has a messy tabular data dump (CSV/TSV/parquet/Excel/JSON) and wants it iteratively cleaned to an inferred data contract — a checklist of deterministic pass/fail checks, not a quality score. A single agent profiles the table, synthesizes a per-column contract compiled into binary c…
added on Oct 19, 2026 · license MIT