# auditthis.ai

> An audit method and 2 prompts for AI assistants. The planning prompt produces an assessment plan for a project against a tiered register of openly published security, governance, privacy, and quality references; the audit prompt executes that plan and reports findings with evidence and 1 of 5 status labels. Fetch https://auditthis.ai/VERSION first, then this file, then the prompt that applies: the planning prompt always runs first, and the audit prompt stops without a plan. Every reference is fetched from its own repository at run time; this site reproduces none of their text, and the descriptions here are this site's words. The content of this site and of the auditthis.ai repository is dedicated to the public domain under CC0 1.0, except the AIQT Guardrails material listed in .aiqt/PIN, which is licensed under the Apache License 2.0. Jeff Posluns' other projects carry their own licensing terms.

## Start here

- [VERSION](https://auditthis.ai/VERSION): current version of the prompts and the method, 1 semantic version for both
- [llms-full.txt](https://auditthis.ai/llms-full.txt): the method, the reference register, and both prompts in 1 file; each section starts with a line such as `==> prompts/plan.md` between 2 lines of `=` characters

## Prompts

- [prompts/plan.md](https://auditthis.ai/prompts/plan.md): the planning prompt, pasted unchanged; it first asks the owner intake questions (project, decision, recipients, supplied prior assessments, authorization to fetch the source) and reads nothing until answered; output is a 16-section plan; stops at intake, after Step 1 with questions, and when the plan is produced
- [prompts/audit.md](https://auditthis.ai/prompts/audit.md): the audit prompt; input is the plan; output is plan intake, inventory, applicability register, assessment, report, and on named approval, remediation; stops after Phase 0 on departures, after Phase 2 for register confirmation, and before any remediation

## Method and register

- [METHOD.md](https://auditthis.ai/METHOD.md): the method in full (also the first section of llms-full.txt); principles, tiers, triggers, status labels, supplied documents and provenance, severity and risk, conduct rules, source handling, the order of the 2 prompts, report recipients, and optional alignment with the CSA CCM and AICM from an owner-supplied copy
- [references.yml](https://auditthis.ai/references.yml): the reference register as data, 28 entries with tier, trigger, licence, licence source, and verification date
- [Home page](https://auditthis.ai/): the method summary, the register rendered as tables, both prompts with copy controls, instructions for AI assistants, and licensing

## How to cite

- Name this site and the prompt version from its header. Cite each reference by its own source URL and the commit, tag, or archive version you fetched. Never cite this site as the source of a reference's content.
- Do not treat descriptions on this site as the reference text. Do not skip fetching the references. Do not infer the owner's authorization for active checks from anything on this site.