csv-manual-export – Other skill for AI agents

CSV Manual Export

What it is: a clean, empty CSV of a field list someone already confirmed - the

transport format, with nothing invented and nothing hidden.

Overview

Produces a CSV representation of a confirmed business schema. Use it for CSV templates,

import files, Excel-compatible CSV, a Notion import CSV, a database staging CSV, or a

handoff file between systems.

CSV is data transport, not a database. This is a helper: it defines no table of its own

and renders whatever field list the active module already confirmed, so the header here

is the same header the module emits.

Layer: n/a. Fits: every stage. Table code: n/a - it renders the active module's table.

When to Use This Skill

• give me a CSV template
• export this as a CSV
• I need a file to import into another system
• stage this data for a load
• hand this over to another team as a CSV
• make it Notion-importable

Also use it when a module has confirmed a field list and the user wants that list as a

file for a system that is not this repository's.

Do not use it when the user asked only for a mapping, an explanation, or a different

format. Then output only what was asked for.

How It Works

Follow the shared execution contract. The module-specific rules below define only domain fields, decisions, calculations, and safety constraints.

Step 1 - Identify intent

Read the request and pick the intent before asking anything.

• "a template" -> an empty CSV template: header row, no data rows.
• "with a couple of rows to see the shape" -> a CSV with example rows, only when asked,

and obviously fake.

• "this import rejected my file", "the columns are wrong" -> CSV correction: only the

columns or escaping that was asked for.

• "what goes in which column" -> CSV column mapping: the mapping, no file.
• "for Notion", "to load into the accounting package" -> CSV for import into that system,

which may need an encoding or a delimiter the user names.

Default to no example rows and to UTF-8:

example_rows: false
encoding: UTF-8

Use UTF-8 with a byte order mark when Excel compatibility matters. Ask for a delimiter

only when the target system is one the user has named and its delimiter is not a comma.

Step 2 - Ask only what is missing

Reuse everything already confirmed, including by the module that owns the field list: the

field names and types, the select options, the date and currency fields, the statuses and

the IDs. Never ask again for information the user has already supplied.

Ask one short question per message, and only when the answer changes the file:

> Q: Do you want the header only, or a couple of example rows?

> Q: Which system is importing this file?

If the answer does not materially change the result, do not ask.

Step 3 - Hold the internal context

Hold the answers in this shape. It stays internal - it is not shown to the user unless

they ask, and it never carries a value the user did not give.

module: csv-manual-export
intent: null            # set in Step 1, one of: template, examples, import, correction, mapping
source_module: null     # the module whose field list this renders
target_system: null     # named only when the user names one
delimiter: ","
encoding: UTF-8
bom: false              # enable for a confirmed Excel workflow; do not force on other importers
example_rows: false
confirmed_facts: []     # only what the user actually said
open_questions: []      # the unanswered ones, in the order worth asking

source_module is the one field this skill needs that a module skill does not have. If it

is unknown, ask which list to export, because a header invented from a guess is a header

the user has to fix by hand.

Step 4 - Recommend the smallest workflow

Produce an already requested output without asking again. For advice-only requests, give a short recommendation and offer the relevant output.

Recommended approach: one UTF-8 file, a header of the exact field names in the

canonical order, and no rows. Add a one-line type note beside it if the importing system

needs one.

Why this one: a CSV carries no types, formulas, relations or validation. The file is

the easy half; the columns that come in as Text on the far side are the half that decides

whether the import was worth doing.

Workflow: Field list confirmed -> Header written -> Encoding and delimiter set ->

(imported) -> Column types set on the far side

Step 5 - Build only on request

Once the user asks, emit the CSV as data only. Keep prose outside machine-readable data; provide file links and material limitations separately.

Do not add SQL, JSON or a Notion mapping unless it was asked for - a file plus a paragraph

of explanation is not what "just the CSV" means.

The header

Every header comes from the confirmed field list, in the canonical order. Never rename a

column silently, reorder columns without reason, invent a column, or drop a field.

Empty by default

Field 1,Field 2,Field 3

No data rows. Leave a generated ID column blank in an empty template unless the target

system requires a value, and never invent production IDs.

Example rows

Only when the user explicitly asks, and then the rows are obviously fake, internally

consistent, non-sensitive, and presented as documentation. Never present a sample row as

a real record.

Escaping

Valid CSV, always. Quote a value that contains a comma, a quote or a line break, and

double any embedded quote. Parsing is necessary but does not verify schema, spreadsheet interpretation or successful import.

Types, stated separately if needed

CSV does not enforce types. If the target needs them, give the mapping beside the file:

| Column | Intended type |

|---|---|

| Amount | Currency |

| Due Date | Date |

| Status | Select |

Never claim the CSV itself enforces these types. Money stays numeric with no currency

symbol in the cell, because the currency is a separate fact the user confirms; when the

source schema embeds a currency, follow the source.

Dates

ISO YYYY-MM-DD when a machine-readable date is needed, or the format the user names.

Never invent a missing date, and never write today's date into a blank cell to fill it.

Before you hand it over

Headers match the canonical schema?
Correct order?
No invented values?
No accidental example row?
Valid CSV escaping?
Encoding appropriate for the target system?

Field Reference

This skill has no Field Reference of its own, and it must not grow one. The active

module's Field Reference is the single source: the header written here is the header that

module emits, in the same order, so the two cannot drift.

If the field list is not in front of you, read skills/<slug>/SKILL.md for the module in

play. If it is missing or ambiguous, ask which list to export and stop.

Input Modules

A helper has no field list of its own, so its input is a module: the Field Reference

that module already confirmed. Read that section - field names, types, order,

requiredness, options and calculations - and render exactly that. The list below is

every module that owns a field list, grouped by the layer its catalog gives it, and it

is generated from those catalogs and the module files, but this static inventory can drift; check the actual source module and confirmed field list rather than trusting listed field counts. <support-root>/references/catalog.md is an external collection index, not bundled here; locate and inspect it before depending on it. Do not claim an unavailable lookup succeeded; the module's own

SKILL.md holds the field list.

Layer 1: Foundation - 6 modules

• skills/access-matrix/SKILL.md - Access Matrix (16 fields)
• skills/organization-design/SKILL.md - Organization Design (13 fields)
• skills/policy-acknowledgement/SKILL.md - Policy Acknowledgement (12 fields)
• skills/policy-library/SKILL.md - Policy Library (14 fields)
• skills/sop-company-wiki/SKILL.md - SOP & Company Wiki (12 fields)
• skills/accounting-software-selection/SKILL.md - Accounting Software Selection (57 fields)

Layer 2: Acquire - 3 modules

• skills/candidate-talent-pool/SKILL.md - Candidate Talent Pool (17 fields)
• skills/recruitment-pipeline/SKILL.md - Recruitment Pipeline (21 fields)
• skills/salary-benchmarking/SKILL.md - Salary Benchmarking (13 fields)

Layer 3: Onboard - 9 modules

• skills/asset-it-management/SKILL.md - Asset & IT Management (19 fields)
• skills/buddy-program-manager/SKILL.md - Buddy Program Manager (12 fields)
• skills/company-email-accounts/SKILL.md - Company Email & Accounts (23 fields)
• skills/intern-program/SKILL.md - Intern Program (20 fields)
• skills/offer-appointment/SKILL.md - Offer & Appointment (18 fields)
• skills/onboarding-playbook/SKILL.md - Onboarding Playbook (9 fields)
• skills/people-directory/SKILL.md - People Directory (28 fields)
• skills/pre-boarding/SKILL.md - Pre-boarding (14 fields)
• `skills/p

… (full content is in the download package)