> ## Documentation Index
> Fetch the complete documentation index at: https://docs.nimt.ai/llms.txt
> Use this file to discover all available pages before exploring further.
# Guided workflows
> Ready-to-use recipes for completing multi-step work with Nimt.
Guided workflows combine several Nimt tools into a reliable sequence. If your
client supports MCP prompts, select the workflow from its prompt menu. Otherwise,
copy the example request into your conversation.
Guided workflows do not start automatically. MCP prompts are selected by you,
and each client decides how to display them. You can always use the copyable
request instead.
## Create a project
Create and configure a Nimt project while reviewing the important decisions
before activation.
> Help me create a Nimt project for example.com in English with global tracking.
> Let me review the tracking models, brand, categories, and prompts before
> activating it.
Replace the website and market with your own. You can also leave them out and
let the assistant ask for the missing information.
### What happens
Nimt helps the assistant:
1. Collect the website and target market.
2. Create a draft and analyze the website.
3. Let you review the tracking models, brand information, and categories.
4. Generate prompts and let you edit them.
5. Summarize the complete project for review.
6. Ask for your confirmation before activation.
Website analysis and prompt generation run in the background. If either is still
running, the assistant reports the current status and can check again later
instead of holding one request open.
### Before activation
The assistant should show the final website, market, tracking models, categories,
active prompt count, and any trial requirement before asking whether to activate.
Review that summary and your client's confirmation before allowing the write.
The MCP prompt is named `create_project_workflow` and appears as **Create a
Nimt project** in clients that expose server prompts. If it is missing,
refresh or reconnect Nimt. Clients without MCP prompt support can use the
example request above with the same project-creation tools.
## Review AI visibility
Get an evidence-backed view of the project's current position, prompt-level
weaknesses, competitors, and citation landscape.
> Review our AI visibility over the latest 30 days. Compare us with the leading
> brands, identify the prompts where we are weakest, and give me three
> evidence-backed priorities. Do not change anything.
The assistant uses one date range and model filter throughout, compares
visibility, share of voice, ranking, and sentiment, and separates retrieved
Nimt data from its interpretation. It does not infer page content or claim that
a cited source caused a result.
The MCP prompt is named `review_ai_visibility` and appears as **Review AI
visibility**.
## Find competitor gaps
Find the tracked prompts where competitors have the clearest advantage and the
sources associated with those prompts.
> Find our most important competitor gaps over the latest 30 days. Show the
> prompt, competing brand, metric difference, trend, and associated cited
> sources for each priority.
The assistant first identifies relevant competitors from the project data,
then compares matching prompts with identical filters. Sources are described as
associated evidence, not as proven causes of competitor performance.
The MCP prompt is named `find_competitor_gaps` and appears as **Find competitor
gaps**.
## Find citation opportunities
Prioritize domains and pages that may deserve further investigation based on
their citation share, movement, category, and affected prompts.
> Find our strongest citation opportunities over the latest 30 days. Prioritize
> them by impact and evidence, and tell me what to investigate next. Do not
> change any classifications.
The assistant reviews both domain- and page-level citation data and can compare
brand performance in answers that cite a shortlisted source. That relationship
is treated as an association, not proof that the source caused the performance.
The MCP prompt is named `find_citation_opportunities` and appears as **Find
citation opportunities**.
## Review prompt coverage
Review whether the tracked prompts cover the project's intents, tags, observed
searches, and important weak spots.
> Review our prompt coverage over the latest 30 days. Find missing themes,
> likely duplicates, untagged prompts, and recurring observed searches that are
> poorly represented. Propose exact changes, but wait for my confirmation
> before applying them.
The assistant inventories the complete prompt and tag catalog before proposing
changes. It can create approved prompts or tags and update approved intent,
active-state, or tag assignments, but it does not replace or delete prompts as
part of this workflow.
Selecting the workflow is not approval to write. Review the complete proposed
change set and explicitly confirm it before the assistant applies anything.
The MCP prompt is named `review_prompt_coverage` and appears as **Review prompt
coverage**.
## Choosing dates and models
The analysis workflows accept an optional project, inclusive start and end
dates, and model. If no dates are supplied, they use the latest 30 days within
the project's available data range and state the exact period. If no model is
supplied, they analyze all models together. Supply a tracked model when you want
a model-specific result.
Continue with the [example questions](/mcp/examples) or review the complete
[tool catalog](/mcp/tools).