SwishrJobs

Return the measured success rate, latency, and expected cost per accepted result for each capability, from controlled synthetic benchmarks. A capability that has not been benchmarked reports measured=false with a null success rate; no rate is ever invented. These are not customer outcomes.

List every task SwishrJobs can execute, with each one's inputs, deliverables, price, and whether it can run right now.

Validate a job request and receive a priced quote without executing anything. Use this to learn the cost and the accepted inputs before authorizing.

Rank capabilities for a described task by task match, measured synthetic success rate, and expected cost per accepted result, and explain the ranking. Use this to decide which capability to run before quoting it. The task may be described in English, Arabic, Spanish, French, German, Portuguese, Chinese, or Japanese; this is discovery only, and every deliverable is produced in English.

Validate an A2A agent card, report every missing field, and deliver a repaired copy. Before publishing an A2A agent card. Do not choose this when you need the agent's endpoints exercised. Example requests: Validate this A2A agent card. | Is my agent card complete for A2A discovery? | Fix the missing fields in this agent card.

Audit an HTML page against the WCAG 2.1 success criteria that can be decided from the markup alone — page language, title, headings, a missing alt attribute, form-field labels, link and button names, table headers, iframe titles and zoom restrictions that stop a screen reader or low-vision user — each failure named with its criterion and conformance level, plus a per-criterion pass/fail table and an overall verdict. Before shipping a page, to catch the WCAG failures a markup scan can decide deterministically. Do not choose this when you need contrast, keyboard or screen-reader testing, which needs a rendered page and a human. Example requests: Audit this HTML page for WCAG 2.1 accessibility failures. | Which WCAG success criteria does this markup fail? | Check this page for missing alt text, labels and a page language. | Run an accessibility conformance audit over this HTML.

Attribute agent usage cost across chosen dimensions and rank the groups. When a usage log needs cost attributed to owners or components. Do not choose this when you need live billing data from a provider. Example requests: Attribute this agent cost across agents and models. | Which agent is spending the most? | Break down this usage log by dimension.

Summarise an agent trace: span count, total and slowest duration, and error spans. When a trace has been exported and needs summarising. Do not choose this when you need live trace data from a backend, which needs access. Example requests: Analyse this agent trace and tell me where the time went. | Summarise these spans and count the errors. | What is slowest in this agent run?

Compare two OpenAPI documents and report every breaking change — removed paths and operations, and newly required parameters — with the affected operation named. Before publishing an API change, to see which operations existing callers lose. Do not choose this when you need a full schema compatibility checker, or when the documents are not OpenAPI JSON. Example requests: Diff these two OpenAPI specs and tell me what breaks for existing clients. | Will upgrading this API version break my integration? Here are both specs. | List the breaking changes between the old and new API contract.

Measure the decay from in-sample to out-of-sample performance and flag parameter-rich strategies. After a backtest, before trusting its in-sample performance. Do not choose this when a full statistical overfitting test is required. Example requests: Check this backtest for overfitting. | How much performance does this strategy lose out of sample? | Flag the overfitting risk in these results.

Match a ledger against a bank statement on id and amount, and list every exception. At month-end, when a ledger must be tied to a bank statement. Do not choose this when entries need fuzzy matching on reference or date. Example requests: Reconcile this ledger against the bank statement. | What does not match between these two files? | Find the breaks between my ledger and statement.

Produce a standard-library Python module that produces boundary values for an integer range, plus pytest tests. Choose this when you need the boundary and just-outside values of an inclusive integer range, with tests. Do not choose this for non-integer ranges, random data generation, or property-based testing frameworks. Example requests: Write a Python module that produces the boundary and just-outside values for an integer range, with tests. | I need boundary values for an inclusive integer range for test data. | Generate a boundary-value module and prove it with pytest.

Generate a runnable argparse command-line tool with a subcommand per declared name, a test module and a README. At the start of a CLI project, to get the argument plumbing and tests in place. Do not choose this when you want the subcommand logic implemented, which needs a specification. Example requests: Scaffold a CLI tool with these subcommands. | Generate an argparse entry point for my new command. | Give me a starter Python CLI with tests and a README.

Check that a service's x402, MCP, A2A and OpenAPI discovery surfaces are all present, well formed and HTTPS-consistent. Before publishing a service for agent discovery across x402, MCP, A2A and OpenAPI. Do not choose this when you need the live endpoints exercised, which needs a request. Example requests: Check my service's x402, MCP, A2A and OpenAPI surfaces for conformance. | Is my service ready for agent discovery? | Are my discovery documents consistent with each other? | Audit my agent commerce surface metadata.

Extract the required contract clauses, map each declared obligation to its supporting sentence, and report the coverage and the gaps. When a contract or policy needs its clause coverage and evidence mapped quickly. Do not choose this when you need legal interpretation or advice, which needs a lawyer. Example requests: Extract the key clauses from this contract and report what is missing. | Map my obligations to the sentences that evidence them. | Is this contract complete for a data-processing review? | Run a compliance evidence check over this agreement.

Prepare an audit-readiness evidence pack from a project's container, infrastructure, CI and dependency artifacts: run the repository hardening and supply-chain scans, map every finding to the SOC 2 Trust Services Criteria and ISO/IEC 27001 Annex A control it bears on, name the controls with no supporting evidence as gaps, and ship a per-finding remediation plan, a control map and a readiness verdict. Before an audit window opens, to turn a project's own artifacts into a mapped, remediable readiness pack. Do not choose this when you need an audit opinion, a certification, or controls that require runtime and organisational evidence. Example requests: Prepare a SOC 2 readiness pack from my repository artifacts. | Map my security findings to SOC 2 and ISO 27001 controls. | Which controls have no evidence in my project? | Build an audit readiness evidence pack for my repository.

Validate 5-field cron expressions, name the field that is out of range, and flag the day-of-month and day-of-week combination that most implementations treat as 'either', so a schedule that never fires is caught. Choose this to validate 5-field cron expressions, to have an out-of-range field named, and to catch the day-of-month plus day-of-week combination that most implementations treat as 'either' and that makes a job never fire. Do not choose this when you need the next fire time or a timezone conversion; it validates field syntax only. Example requests: Check these cron expressions for mistakes. | Why does this cron job never run? Here are the expressions. | Validate my cron scheduled jobs before I deploy them.

Clean a CSV and return the cleaned file plus a record of every change. Trims cells, makes duplicate and empty column names unique, drops blank and duplicate rows, and pads or truncates ragged rows. Choose this when you want the CSV itself tidied and a change log: cells trimmed, column names made unique, blank and duplicate rows removed, ragged rows padded or truncated. Do not choose this when you only need the problems listed (use csv-preflight), or when deciding which duplicate to keep needs a business rule: this keeps the first occurrence. Example requests: Clean this CSV: trim the fields, drop the blank lines and remove duplicate rows. | Normalise this export before I load it, and tell me what you changed. | Fix the duplicate column names and ragged rows in this CSV.

Profile every column of a CSV: inferred type (number, integer, date, email or text), non-empty and missing counts, distinct values, and min/max for numeric columns. Choose this for a per-column picture of a CSV: inferred type, non-empty and missing counts, distinct values, and min/max for numeric columns. Do not choose this when you want the file diagnosed for import errors (use csv-preflight) or a completeness report (use csv-missing-value-report). Example requests: Profile the columns of this CSV before I load it. | Profile which CSV column is numeric and what is its range? | How many distinct values are in each CSV column? Profile this file.

Detect date columns in a CSV, report the formats present, flag dates that could be read as day/month or month/day, and return the file with every date rewritten as ISO 8601. Choose this before loading a CSV whose dates are in mixed or local formats, to detect the date columns, flag dates that could be read as day/month or month/day, and receive the file rewritten as ISO 8601. Do not choose this when dates carry times or timezones, which are not converted, or for two-digit years. Example requests: Normalise the dates in this CSV to ISO format. | Are any of these dates ambiguous? | Convert these UK-style dates to ISO 8601.

Find duplicate rows in a CSV, group each set with the row numbers it appears on, and return a deduplicated file keeping the first occurrence. Choose this when a CSV has repeated rows and you need them collapsed and named: the report groups each duplicate set with its row numbers and the delivered file keeps one copy of each row. Do not choose this when duplicates must be judged on a specific key column rather than the whole trimmed row. Example requests: Remove duplicate rows from this CSV and tell me which were duplicates. | How many duplicates are in this CSV file? | Give me a deduplicated copy of this CSV.

Describe how a CSV is actually shaped: the detected delimiter, the column count, ragged rows with their numbers, duplicate rows, and columns that are empty in every row. Choose this when you need to understand a file's shape — its delimiter, column count, ragged rows, duplicate rows and fully empty columns — before parsing it. Do not choose this when you want the file cleaned (use csv-clean), import errors diagnosed (use csv-preflight) or per-column statistics (use csv-column-profile). Example requests: What delimiter is this CSV actually using? | Report the shape of this CSV before I load it. | Why does this CSV import with the wrong number of columns?

Produce a standard-library Python module that transforms CSV text into a JSON array of objects, plus pytest tests. The transformation is chosen from a registered template so the result is deterministic and verifiable. Choose this when you need a tested, standard-library function that parses CSV text into a JSON array of objects, one per row. Do not choose this for per-column type coercion or bespoke rules, or if you need the file converted for you rather than a module that does it: this delivers code, not converted data. Example requests: Give me a Python module that turns CSV text into a JSON array of objects, with tests. | I have CSV and need JSON objects, one per row, produced by a tested function. | Convert CSV to JSON and prove it with a pytest suite.

Count how much is missing in each column of a CSV, separating truly blank cells from sentinel values such as NA, null, none, '-', and reporting present counts. Choose this before analysis to see how complete each column is, with blank cells and sentinel values such as NA, null, none and '-' counted separately. Do not choose this when you need values imputed or a custom sentinel recognised; only the built-in sentinel list is used. Example requests: Which columns of this CSV have the most missing values? | How many missing, NA and blank cells are in this CSV? | Report completeness per column before I analyse this file.

Check a CSV before it is imported. Reports ragged rows, duplicate and empty column names, blank and duplicate rows, per-column type drift and column profiles, each with the row number a spreadsheet would show. Choose this when you have a CSV file and want the problems listed with the row numbers a spreadsheet shows: ragged rows, duplicate or empty column names, blank and duplicate rows, and per-column type drift. Do not choose this when you want the file fixed rather than diagnosed (use csv-clean), when you want a module that validates records in your own pipeline (use data-validation-module), or for a binary workbook such as xlsx, which this service cannot read. Example requests: Check this CSV for problems before I import it into Postgres. | Why does my CSV import fail? Here is the file. | Preflight a customer data export and tell me which rows are malformed.

Score a CSV on completeness, type consistency and uniqueness, per column, and report the columns that are always empty, mostly empty, constant, or hold mixed types, plus duplicate rows. Choose this when you need a quick, reproducible quality assessment of a dataset: a score plus per-column completeness, uniqueness and type consistency. Do not choose this when you need the specific malformed rows listed (use csv-preflight), or when you need a module that produces a quality report inside your pipeline (use dataset-quality-report). Example requests: Profile this CSV and tell me which columns are unreliable. | Score this dataset export for quality before I trust it. | Which columns in this CSV are mostly empty or always the same value?

Convert a CSV into a JSON array of objects keyed by the header, with documented type coercion: empty becomes null, integers and numbers become numbers, true and false become booleans, everything else stays a string. Choose this when you need a CSV delivered as JSON records, with documented type coercion: empty to null, numbers to numbers, true and false to booleans. Do not choose this when you need a module that performs the conversion inside your own pipeline (use csv-json-transformation), or when the target shape is nested: CSV is flat. Example requests: Turn this CSV into a JSON array of objects. | Convert this spreadsheet export to JSON for my API. | Give me this CSV as JSON records, with numbers as numbers.

Turn a list of business questions into a dashboard specification with a tile and visualisation per question. At dashboard design time, before building it in a BI tool. Do not choose this when you need the dashboard built or the data queried. Example requests: Turn these questions into a dashboard spec. | Design a dashboard for these business questions. | What tiles should this dashboard have?

Derive a table-level lineage graph from a set of create/select statements, with a Mermaid diagram. When a set of SQL models needs a dependency graph. Do not choose this when column-level or runtime lineage is required. Example requests: Build a lineage graph from these SQL files. | What depends on what across these models? | Show the data lineage for these tables.

Profile a source extract against a target schema: per-column type and null profile, unmapped columns, absent target fields, and a go/no-go readiness verdict. Before a data migration or a bulk load, to find the blockers and the weak columns. Do not choose this when you need the data transformed or loaded, which needs the systems. Example requests: Is this data ready for migration to my target schema? | Profile this extract and tell me what blocks the migration. | Which columns are unmapped or mostly empty before a data migration? | Run a data migration readiness check on this CSV and target schema.

Produce a standard-library Python module that validates a record and returns structured errors, plus pytest tests. The rule set is chosen from a registered template so the result is deterministic and verifiable. Choose this when a record must be checked against a registered rule set and the result is a list of human-readable errors, with tests. Do not choose this for rules outside the registered template, or when you need a schema-validation library rather than a self-contained module. Example requests: Write a Python module that validates a record against a fixed rule set and returns a list of error strings. | I need a validation function that returns an empty list for valid input, with tests. | Generate a data-validation module and its pytest suite.

Review a SQL migration for destructive and blocking operations — dropped tables and columns, NOT NULL columns without defaults, type changes and renames — and deliver a safe deployment order. Before running a migration against a database that is serving traffic. Do not choose this when you need an execution plan or an actual lock-time estimate. Example requests: Is this migration safe to run on a live database? | Review this SQL migration for downtime risks. | What could go wrong if I run this migration?

Produce a standard-library Python module that computes a quality report for an array of records, plus pytest tests. Choose this when you need a computed quality summary of a record array: counts, presence, nulls, completeness ratio, exact duplicates. Do not choose this for statistical or semantic quality (outliers, drift, schema conformance beyond the fixed checks). Example requests: Write a Python module that reports record count, per-field presence and duplicate count for a dataset, with tests. | I need a data-quality summary: null counts, completeness ratio, exact duplicates. | Generate a dataset-quality report module and test it.

Parse a dbt model's SQL and generate a schema.yml with not_null tests for every selected column. When a dbt model needs baseline tests generated from its SQL. Do not choose this when the model uses jinja or macros a textual parse cannot resolve. Example requests: Generate dbt tests for this model. | Write a schema.yml for my dbt model. | Add not_null tests to this dbt model's columns.

Match a lockfile against an advisory feed and return the affected dependencies ranked by severity. When you have a lockfile and an advisory list and want the overlap ranked. Do not choose this when you need live advisory data or transitive resolution (use a scanner). Example requests: Which of my dependencies have known vulnerabilities? | Triage this lockfile against the advisory feed and rank the risk. | Are any of these pinned packages affected by a CVE?

Infer per-column data-quality rules from a sample and generate a test module plus a rules file. When a new dataset needs baseline quality checks. Do not choose this when rules must be authored against a specification, not inferred. Example requests: Generate data-quality rules from this sample. | What checks should I run on this dataset? | Create a rules file and tests for this data.

Produce a standard-library Python module that groups records sharing a key field, plus pytest tests. Choose this when you need each repeated key value reported with its count and the positions of its members. Do not choose this to remove duplicates (use record-deduplication) or to match near-duplicates. Example requests: Write a Python module that groups records by a key field and reports duplicated values, with tests. | I need to know which key values occur more than once, with counts and positions. | Generate a duplicate-group report module and prove it with pytest.