qredence/fleet-rlm/tabular-analysis

Claude Code and agent skill by qredence/fleet-rlm

Trust tier 3
  • community
Source check pending

Checking that the SKILL.md file is still in the repository.

Aggregate, filter, join, and compare CSV or JSON data in /workspace (totals, rates, group-bys, quarter or period comparisons, counterfactuals), write a JSON report and a verify script under artifacts/, and answer with figures taken from executed code.

Install

npx vskill@latest install qredence/fleet-rlm/tabular-analysis

SKILL.md

Tabular analysis

Use this for questions answered by computing over rows: sums, averages, rates, group-bys, rankings, period-over-period changes, and "what if" recomputations.

Use the snapshot's installed Python packages and the task project's own dependency manifest when one applies. Do not install packages into the shared snapshot during a task. If a required dependency is unavailable, explain the limitation or ask whether an isolated optional environment is acceptable.

Steps

  1. Inspect before computing. Print the header, the row count, a few rows, and

the distinct values of the columns you will group or filter on:

```python import csv

rows = list(csv.DictReader(open("/workspace/files/sales.csv", newline=""))) print(len(rows), list(rows[0])) print(sorted({r["status"] for r in rows})) ```

  1. Convert types explicitly (int(...), float(...)) and state every filter in

code, for example status == "ok". Keep the result in variables such as totals, rates or report.

  1. Print every figure you will report. Round only for display, for example

round(x, 2) for money and round(x, 4) for rates.

  1. Write results with workspace.write_artifact("report.json", json.dumps(report, indent=2, sort_keys=True)).

It returns the path, which is relative to artifacts/.

  1. When the task asks for verification, write an independent check (another

filter path, or a standalone verify_report.py that reloads the CSV and compares it with the JSON), run it with subprocess.run(["python", path]), and require that it prints OK.

  1. SUBMIT from variables, never from retyped numbers:

``python answer = f"Highest revenue: {top} (${totals[top]:,.2f})." SUBMIT(answer=answer, learned_skill_note=note) ``

Fleet rejects a SUBMIT whose figures were typed into the code but never printed by an execution.

Pitfalls

  • Rates need an explicit denominator ("refunded / all orders in the region,

any status" is not the same as "refunded / ok orders").

  • Quarters are (month - 1) // 3 + 1.
  • For a counterfactual ("if refunded orders had been ok"), add the refunded

revenue to the computed base. Do not reuse a total from another filter.

  • Prior-turn reports are under artifacts/. If one is missing in this

sandbox, recompute from files/ instead of recalling numbers.

Security

  • Security scanPassed, 90 of 100Pattern checks on the skill's files.
  • Model reviewPassed

Quality evaluation

  • Task evaluationNot runNo evaluation yet. Run one in Skill Studio.

Source

Popularity

7-day trend score
0

Not refreshed yet.

Works with

Any agent vskill installs to: 56 targets in all. These 39 have a page on this site.

  • Claude Code
  • Cursor
  • GitHub Copilot
  • Windsurf
  • Codex
  • Gemini CLI
  • Cline
  • Amp
  • Roo Code
  • Goose
  • Continue
  • Kilo Code
  • Junie
  • Trae
  • OpenHands
  • Qwen Code
  • AI Agent
  • Tabnine Chat
  • Sourcegraph Cody
  • Amazon Q
  • Firebase Genkit
  • Sweep
  • Devin
  • Phind
  • Mentat
  • Claude Desktop
  • Aider
  • GPT Pilot
  • Plandex
  • bolt.new
  • v0
  • Replit Agent
  • Anse
  • TypingMind
  • BetterChatGPT
  • LibreChat
  • Warp AI
  • Pieces
  • Fig AI
Embed badge
vskill VERIFIED badge
[![vskill VERIFIED](https://verified-skill.com/api/v1/skills/qredence/fleet-rlm/tabular-analysis/badge)](https://verified-skill.com/skills/qredence/fleet-rlm/tabular-analysis)

Something wrong with this skill or its page? Tell us.