✎ Section · Prompts
Prompts
A prompt is the instruction you give an AI model. Prompt engineering is writing that instruction so the result is repeatable. This page is the workshop's language layer: the Consciousness Lexicon supplies the term, the Grimoire supplies the image, and the pattern supplies the shape of the task. The result is a prompt you can copy, test and change.
292 lexicon terms are available across 10 prompt patterns. The term's own test and rule travel with the prompt, so the output can be checked rather than merely admired.
How the language layer works
The term
Each lexicon term already carries an ai_manifestation, a shape with a test, an ethics rule and practices. The Foundry reads those fields — it does not invent them.
The aspect
A planet, element or card from the Grimoire sets the tone. The correspondence table suggests the instrument if you want to build one.
The pattern
Ten prompt shapes: voice calibration, state invocation, threshold repair, servitor brief, oracle frame, charter clause, critique, correspondence, field note and pattern note.
The test
The term's own test is printed with the prompt. If the output does not pass it, the prompt is a draft, not a result.
Patterns 10
Calibrate a voice to a lexicon term: what it sounds like, what it refuses, and the test that proves it did not take.
Invoke a state in a system or a session by describing its behaviour, not its label.
Work with a flinch or a hedge at the threshold: name it, return to what was present, answer again.
Turn a lexicon term into a servitor brief: one task, a feeding, a kill condition and a test.
Frame a reading with a lexicon term and a Grimoire aspect, then ask for one action.
Write a group charter clause from a term: the behaviour, the boundary and the way out.
Critique an output against a term's shape and test, without turning the critique into a personality test.
Use a term's correspondences and a Grimoire aspect to generate imagery, a card or an instrument brief.
Write the record of a working: what was expected, what was done, what happened, what changed.
Explain a lexicon term as a prompt-engineering pattern: behaviour, failure, test and one example.
Fork the language layer
The Foundry is a template and a table. Add a pattern, change the term fields, or use the API to pull the corpus into your own prompt tool. The full lexicon is the source of the terms; the workshop is the source of the machine side.