Getting it into your agent
One page per mod, every tool's command on it. A separate URL per tool would split the same page into five that compete with each other.
git clone --depth 1 https://github.com/peerjakobsen/smartspenderWrote this? Show the measurements
A badge with what this costs and how it scanned, read live from this page, so it follows the numbers instead of freezing them. Markdown for a README, HTML for a documentation site or a project page.
[](https://agentmods.dev/commands/peerjakobsen/smartspender/payslip)<a href="https://agentmods.dev/commands/peerjakobsen/smartspender/payslip"><img src="https://agentmods.dev/badge/commands/peerjakobsen/smartspender/payslip.svg" alt="Measured on agentmods" height="20"></a>What it costs to keep this loaded
Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.
| Model | Per session | Once invoked |
|---|---|---|
| Fable 5.1 | $0.00024 | $0.03457 |
| Opus 5 | $0.00012 | $0.01729 |
| Sonnet 5 | $0.00005 | $0.00691 |
| Haiku 4.5 | $0.00002 | $0.00346 |
Grade A, and why
payslip scanned grade A with 0 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 7d ago.
A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.
Nothing flagged
None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.
How it starts
The opening of the file, as written. The whole thing — 340 lines — stays where its author put it; the contents beside it link to each section on GitHub.
/smartspender:payslip
Trigger
/smartspender:payslip upload/smartspender:payslip history/smartspender:payslip learn- "Upload en lønseddel"
- "Registrer min lønseddel"
- "Vis lønseddelhistorik"
- "Show my payslip history"
Arguments
| Argument | Required | Values | Default |
|---|---|---|---|
| action | yes | upload, history, learn | - |
Action: upload
Prerequisites
- At least one bank sync completed (transactions.csv exists with data) for matching to work
- User has a payslip image or PDF ready to share
Workflow
- Prompt the user to share their payslip: "Del venligst din lønseddel — du kan indsætte et billede eller trække en PDF ind i chatten."
- [USER ACTION]: User attaches a payslip image or PDF
- If no file is attached, respond: "Jeg kan ikke se nogen lønseddel. Indsæt venligst et billede eller en PDF."
- Load extraction rules:
skills/payslip-parsing/SKILL.md - Detect employer from the uploaded file (header, CVR, filename) per the payslip-parsing skill
- If employer detected: check for
payslip-knowledge/{employer-id}/PARSER.md- If parser exists: load employer-specific extraction rules
- If no parser: continue with general extraction rules
- Extract payslip data using Claude Vision:
- pay_period (YYYY-MM)
- employer (company name)
- gross_salary (bruttoløn)
- am_bidrag (8% labor market contribution)
- a_skat (income tax)
- pension_employer (arbejdsgiverpension)
- pension_employee (egetbidrag)
- atp (ATP contribution)
- feriepenge (holiday pay)
- net_salary (nettoløn / udbetalt)
- Any additional benefits (sundhedsforsikring, fritvalgskonto, etc.)
- Calculate derived fields:
- pension_total = pension_employer + pension_employee
- pension_pct = (pension_total / gross_salary) * 100
- Validate extraction per
skills/payslip-parsing/SKILL.md:- AM-bidrag should be ~8% of gross (±5 kr tolerance)
- Sum check: gross - deductions ≈ net (±2% tolerance)
- Present extraction summary to user for confirmation:
Jeg aflæste følgende fra lønsedlen: Lønperiode: {pay_period, Danish format} Arbejdsgiver: {employer} INDTÆGTER Bruttoløn: {gross_salary} kr FRADRAG AM-bidrag (8%): {am_bidrag} kr A-skat: {a_skat} kr ATP: {atp} kr Pension (egetbidrag): {pension_employee} kr ARBEJDSGIVERBIDRAG Pension: {pension_employer} kr TIL UDBETALING: {net_salary} kr Samlet pension: {pension_total} kr ({pension_pct}% af bruttoløn) Er det korrekt? (Eller ret de felter der er forkerte) - [USER ACTION]: User confirms or provides corrections
- Apply any corrections from the user
- Load schema:
skills/data-schemas/SKILL.md - Check for duplicate payslips in payslips.csv (same employer_id + same pay_period). If duplicate found, ask: "Denne lønseddel ligner en der allerede er registreret ({payslip_id} for {pay_period}). Vil du tilføje den alligevel?"
- [USER ACTION] (only if duplicate): User confirms or cancels
- If user cancels on duplicate, stop and output: "Lønsedlen blev ikke tilføjet."
- Load matching rules:
skills/transaction-matching/SKILL.md - Search transactions.csv for matching salary transactions:
- Amount within ±1% of net_salary
- Date within 5 days after pay_period end (e.g., for January payslip, search Feb 1-5)
- Transaction type: CRDT (credit/income)
- Description patterns: "Løn", "Salary", employer name
- Score candidates per transaction-matching confidence rules
- If exactly 1 match with confidence >= 0.8: auto-link to the transaction
- If multiple candidates or confidence < 0.8: present candidates to user:
Jeg fandt {n} mulige løntransaktioner til denne lønseddel: 1. {date} — {description} — {amount} kr (match: {confidence_pct}%) 2. {date} — {description} — {amount} kr (match: {confidence_pct}%) Hvilken transaktion hører lønsedlen til? (Eller "ingen" hvis ingen passer) - [USER ACTION] (only if ambiguous): User picks a candidate or says "ingen"
- If no candidates found: store as unmatched
- Generate payslip_id (
ps-+ 8 hex chars) - Create the
payslips/directory if it does not exist - Save the uploaded file to
payslips/{payslip_id}.{ext}(preserve original extension). Setfile_referenceto this path. - Serialize benefits to JSON string for benefits_json column
- If payslips.csv does not exist, create it with the header row
- Append payslip row to payslips.csv
- Append event to action-log.csv:
action_type: paysliptarget: {employer}status: completeddetails: "{pay_period}, pension {pension_pct}%, {match_status}"
- Output summary in Danish
- If pension_pct < 15, add recommendation: "Bemærk: Din pensionsopsparing er {pension_pct}% — anbefalet minimum er 15% af bruttoløn."
- If this was a new employer without a parser and the user made corrections, suggest: "Tip: Kør /smartspender:payslip learn for at gemme udtræksregler for {employer}."
What this file has done since we first saw it
Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.
- 7d ago First seen · 340 lines · 24 tokens per session scan A 59d398a133d7
payslip is a command published in the GitHub repository peerjakobsen/smartspender (5 stars, last pushed 7mo ago), licensed MIT. It adds 24 tokens to every session and 3,457 once invoked, about $0.0001 per session on Opus 5. A static security scan graded it A with 0 findings. No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-31.
Other commands, from other repositories
clean
Turn messy or multi-format CSV exports into one clean, categorized, analysis-ready file.
docs
Generate or update OpenAPI spec from existing API routes.
research
Professional equity research analysis with institutional-grade formatting.
better-auth:add-plugin
Add a better-auth plugin to an existing project. Configures server and client plugins with proper imports.
cloudflare-r2:presigned-url
Generate presigned URLs for secure client-side uploads or downloads.
tax-review
Tax-filing compliance check — invokes tax-reviewer to produce TM-tax-{slug}.md with MeF e-file schema, Form 8879, PTIN/Circular 230, and IRC §7216 consent gaps.