Thank you for considering a contribution. This library exists because GRC practitioners share what works.
A new prompt is worth adding if:
- It addresses a GRC task that practitioners perform regularly
- It does not duplicate an existing prompt
- The output would be difficult to produce without AI assistance, or significantly faster with it
- It follows the prompt structure below
An improvement to an existing prompt is worth submitting if:
- You used the prompt in real work and found a specific failure mode
- The fix makes the output more reliable or more specific
- You have tested the improvement and can show a before/after
Every prompt must follow this format:
<role>
[Specific senior expert role with domain and experience credentials]
</role>
<context>
[VARIABLE_ONE] - What this variable represents
[VARIABLE_TWO] - What this variable represents
</context>
<instructions>
Step 1: [First action - always a diagnostic or analysis step before producing content]
Step 2: [Main production step]
Step 3: [Output refinement or summary]
</instructions>
<output_format>
[Specific format instructions]
Write in British English. No em dashes.
CRITICAL OUTPUT RULE: Never use em dashes (the — character) anywhere in your response. Replace every em dash with a comma, a colon, or rewrite the sentence. This rule has no exceptions.
</output_format>
Non-negotiable rules for all prompts:
- British English throughout
- No em dashes in the prompt text (except the one in the CRITICAL OUTPUT RULE showing the character to avoid)
- Step 1 must be a diagnostic or calibration step, not immediate production
- All variables in
[SQUARE_BRACKETS]with a description <output_format>must include the CRITICAL OUTPUT RULE verbatim
- Fork the repository
- Create a branch:
git checkout -b add-prompt-NAMEorgit checkout -b fix-prompt-NUMBER - Add or edit the prompt file in the appropriate domain folder
- Update the prompt index in
README.mdif adding a new prompt - Open a pull request with:
- What the prompt does
- Which GRC domain it belongs to
- The complexity level (Essential / Practitioner / Advanced) and why
- If an improvement: what failure mode you found and how the fix addresses it
- Generic prompts that could apply to any profession (not GRC-specific)
- Prompts that duplicate existing functionality without meaningful improvement
- Prompts without proper
<role>,<context>,<instructions>,<output_format>structure - Prompts that produce outputs you have not tested
- Anything with em dashes in the prompt text
Open an issue if:
- A prompt produces unreliable or incorrect outputs for a specific use case
- A prompt is missing a constraint that would make it safer or more accurate
- A GRC use case is not covered and you want to discuss it before building
Be direct. Be specific. Be useful. GRC practitioners are busy people. Contributions that waste their time are not welcome. Contributions that save their time always are.