Data Analysis

Indexed Reference Map

• @[SKILL.md#L1-L64] Frontmatter, scope, KB notes, trigger rules, SOP, error handling, output contract.
• @[references/INDEX.md#L1-L11] Reading order and document map.
• @[references/QUICK_REFERENCE.md#L1-L22] Fast decision map and execution ladder.
• @[references/CORE_API.md#L1-L44] Action contract, file/path rules, cache policy, and dependency expectations.
• @[references/PATTERNS.md#L1-L26] Query patterns, schema-first discipline, and analysis scope control.
• @[references/EXAMPLES.md#L1-L18] Example asks and expected workflows.
• @[references/AI_WORKFLOW.md#L1-L14] Bounded execution sequence.
• @[scripts/analyze.py#L1-L350] Local-safe analysis helper.

Scope

• In-scope:
  • Inspecting CSV/XLSX/XLS structure and schema.
  • Running analytical SQL against one or more tabular files.
  • Producing summaries and exporting result sets.
• Out-of-scope:
  • Blind execution against unknown files without first understanding schema and user goal.
  • Auto-installing dependencies or fetching extensions at runtime.
  • Treating arbitrary paths outside the approved workspace boundary as default inputs.

KB Mode Notes

• Reading order:
  1. references/INDEX.md
  2. references/QUICK_REFERENCE.md
  3. references/CORE_API.md
  4. references/PATTERNS.md
  5. references/EXAMPLES.md
  6. references/AI_WORKFLOW.md
• Coverage Matrix:
  • trigger/use-vs-don't-use -> QUICK_REFERENCE.md
  • actions, paths, cache, exports, dependencies -> CORE_API.md
  • querying/inspection patterns -> PATTERNS.md
  • representative tasks -> EXAMPLES.md
  • bounded execution discipline -> AI_WORKFLOW.md
• Portability note:
  • Workspace pathing and local dependency assumptions here are source-of-truth for OpenClaw-Module Android; verify before porting elsewhere.

Trigger Rules

• Positive:
  • User provides or references CSV/Excel files and wants analysis, summary, filtering, joins, or exports.
  • The task benefits from SQL-based exploration or structured summaries.
• Negative:
  • The task is plain-text document summarization rather than tabular data analysis.
  • Required dependencies are missing and the user has not asked to repair the Python environment.
• Escalation:
  • Start with schema inspection before writing non-trivial SQL.
  • If file paths are outside the safe workspace boundary, pause for approval or relocation into the workspace.

SOP

1. Confirm the input files and the analytical question.
2. Run inspect first to discover sheet/table names, columns, and sample rows.
3. Build the minimum SQL or summary action needed for the question.
4. Export only when the user needs a file artifact; otherwise present concise findings directly.
5. Keep outputs and cache artifacts in approved workspace-safe paths.
6. State assumptions, table-name mappings, and limitations when relevant.

Error Handling

• Missing dependency (duckdb, openpyxl):
  • stop and report the missing package; do not auto-install from inside the helper.
• Unsupported file/path:
  • return a bounded error and list accepted formats/expected locations.
• SQL failure:
  • show the error plus available tables/columns for recovery.
• Oversized/broad ask:
  • inspect schema first, then narrow the question before deeper analysis.

Output Contract

• Artifact:
  • direct findings in chat and/or exported .csv, .json, or .md results.
• Validation:
  • schema inspection precedes complex analysis.
  • path handling respects workspace boundaries and .ops/* taxonomy.
  • helper performs no hidden dependency install or network extension fetch.
• Compliance Verdict:
  • PASS when analysis flow is schema-first, workspace-safe, and reproducible.