Today I'm releasing whenwords, a relative time formatting library that contains no code.

whenwords delivers the following functions:

• timeago — Converts a timestamp to a relative time string like "3 hours ago" or "in 2 days" based on a reference time.
• duration — Formats a number of seconds as a human-readable duration like "2 hours, 30 minutes" or "2h 30m" in compact mode.
• parse_duration — The inverse of duration—parses strings like "2h 30m" or "2 hours and 30 minutes" into seconds.
• human_date — Returns contextual date strings like "Today", "Yesterday", "Last Tuesday", or "March 5" depending on how far the date is from a reference point.
• date_range — Formats a start and end timestamp as a smart date range, collapsing redundant information: "March 5–7, 2024" instead of "March 5, 2024 – March 7, 2024".

There are many libraries that perform similar functions. But none of them are language agnostic.

whenwords supports Ruby, Python, Rust, Elixir, Swift, PHP, and Bash. I'm sure it works in other languages, too. Those are just the languages I've tried and tested.

(I even implemented it as Excel formulas. Though that one requires a bit of work to install.)

But like said: the whenwords library contains no code. Instead, whenwords contains specs and tests, specifically:

• SPEC.md: A detailed description of how the library should behave and how it should be implemented.
• tests.yaml: A list of language-agnostic test cases, defined as input/output pairs, that any implementation must pass.
• INSTALL.md: Instructions for building whenwords, for you, the human.

The installation instructions are comically simple, just a prompt to paste into Claude, Codex, Cursor, whatever. It's short enough to print here in its entirety:

Implement the whenwords library in [LANGUAGE].

1. Read SPEC.md for complete behavior specification
2. Parse tests.yaml and generate a test file
3. Implement all five functions: timeago, duration, parse_duration,
   human_date, date_range
4. Run tests until all pass
5. Place implementation in [LOCATION]

All tests.yaml test cases must pass. See SPEC.md "Testing" section
for test generation examples.

Pick your language, pick your location, copy, paste, and go.

Idris2 library implementation by DeepSeek 3.1 Terminus via Crush AI

Prompt:

You have following bash tools:
1. System compiler:
/root/.idris2/bin/idris2 --check --log 50 Sample.idr
2. Ask compiler about hole context, where hole is a "?hole" in code:
echo -e ':color off\n :load "Sample.idr"\n :ti hole' | /root/.idris2/bin/idris2
3. Ask compiler about type declarations, where some_function in code is a top level function or data:
echo -e ':color off\n :load "Sample.idr"\n :di some_function' | /root/.idris2/bin/idris2

Implement Idris2 library according to specification. By the end of implementation run tests to make sure that all is fine and fix code if needed.

Full implementation required to repeat the prompt few times:

run tests to make sure that all is fine and fix code if needed

Overall generation took ~3 hours and spent ~3.5$ at OpenRouter.