Tool Reference

Complete documentation for all Cutline MCP tools available through the Model Context Protocol.

Audit Tools

These tools scan your codebase for active issues — real vulnerabilities and anti-patterns in your code as written today.

engineering_audit

🔓 FREE — Engineering audit that evaluates your codebase against a stack-aware constraint graph covering security, reliability, and scalability. No deep dive or product_id required. 3 scans/month on free tier.

Parameters:

Parameter	Type	Required	Description
project_root	string	Yes	Absolute path to the workspace root
task_description	string	No	Context about what you're building
max_file_bytes	number	No	Max bytes per file (default: 50000)

Response: Readiness scores (engineering, security, reliability, scalability), binding coverage, top finding with fix guidance, and SCA vulnerability summary.

---

code_audit

🔒 PREMIUM — Security-focused code audit that cross-references the product constraint graph, runs LLM extraction for security gaps, and builds an RGR remediation plan.

Parameters:

Parameter	Type	Required	Description
product_id	string	Yes	Product ID from .cutline/config.json
project_root	string	Yes	Absolute path to workspace root
file_paths	string[]	No	Specific files to audit
task_description	string	No	What you're building
audit_mode	string	No	security (default) or full
max_file_bytes	number	No	Max bytes per file (default: 50000)
auto_start	boolean	No	Queue a deep dive job (default: true)

Response: RGR plan, security gaps, extracted risks, checklist summary, SCA findings, constraint graph updates, and next-step RGR instructions.

---

Constraint Tools

These tools are available through the cutline-constraints server. They power the constraint graph, NFR boundary enforcement, and the adaptive Red-Green-Refactor workflow.

rgr_plan

Assess constraint complexity for a file and return an adaptive execution plan. Call this before implementing any feature.

Parameters:

Parameter	Type	Required	Description
product_id	string	Yes	Product ID
file_path	string	Yes	File path to assess (e.g., src/api/upload.ts)
depth	number	No	BFS traversal depth (default: 2)
auth_token	string	No	Firebase ID token

Response: Returns strategy ("full" or "phased"), an array of phases with names, categories, and constraint counts, plus a rationale string.

---

graph_get_boundaries

Get constraint boundaries for a specific file path. Returns functional context (features, components, data types) and applicable NFRs. Supports RGR phased delivery.

Parameters:

Parameter	Type	Required	Description
product_id	string	Yes	Product ID
file_path	string	Yes	File path to look up constraints for
depth	number	No	BFS traversal depth (default: 2)
max_tokens	number	No	Token budget for output (default: 2000)
phase	string	No	RGR phase filter: test_spec, functional, security, performance, economics, full, or auto (default: full)
auth_token	string	No	Firebase ID token

Response: Imperative Markdown with functional context and NFRs, formatted for the requested phase.

---

constraints_auto

Proactive constraint injection. Analyzes your current file/code context and automatically surfaces relevant constraints.

Parameters:

Parameter	Type	Required	Description
product_id	string	Yes	Product ID
file_paths	string[]	No	Files you're working on
code_snippet	string	No	Current code context (500 char max)
task_description	string	No	What you're trying to accomplish
mode	string	No	Output mode: silent, advisory, blocking, deep, auto (default: auto)
max_constraints	number	No	Max constraints to return (default: 5)
phase	string	No	RGR phase filter (same options as graph_get_boundaries)
auth_token	string	No	Firebase ID token

---

rgr_complete_phase

🔒 PREMIUM — Mark an RGR phase as complete. Updates the constraint graph and recomputes readiness scores with the phase bonus.

Parameters:

Parameter	Type	Required	Description
product_id	string	Yes	Product ID
phase	string	Yes	Phase to complete: test_spec, functional, security, performance, or economics

Response: Updated readiness scores and completed phases list.

---

generate_cutline_md

🔒 PREMIUM — Generate and write Cutline config files. Creates .cutline.md (constraint routing engine), Cursor rules, Claude Code config, and RGR workflow rules. If project_root is provided, files are written directly to disk.

Parameters:

Parameter	Type	Required	Description
product_id	string	Yes	Product ID
product_name	string	No	Human-readable product name
project_root	string	No	Workspace root — writes files directly when provided
project_type	string	No	Primary framework/stack (auto-detected if omitted)
mcp_server	string	No	MCP server name to reference (default: cutline-constraints)
max_embedded_constraints	number	No	Max critical constraints to embed (default: 7)

Response: When project_root is set, returns list of files written. Otherwise returns content for manual saving.

---

Pre-mortem Tools

These tools are available through the cutline-premortem server.

premortem.run

Generate a complete pre-mortem analysis synchronously.

Parameters:

Parameter	Type	Required	Description
input	object	Yes	Pre-mortem input configuration
input.productName	string	Yes	Name of the product or feature
input.productDescription	string	Yes	2-3 sentence description
input.referenceClasses	string[]	No	Industry/domain references
input.pessimism	number	No	Pessimism level (0.0 - 1.0, default: 0.5)
auth_token	string	No	Firebase ID token (auto-filled by CLI)

Example:

{
  "input": {
    "productName": "Mobile Analytics Dashboard",
    "productDescription": "A mobile app that provides real-time analytics for e-commerce stores, helping merchants track sales and customer behavior on the go.",
    "referenceClasses": ["Mobile apps", "Analytics tools", "E-commerce"],
    "pessimism": 0.6
  }
}

Response: Full pre-mortem document with assumptions, risks, experiments, and recommendations.

---

premortem.list

List all pre-mortems for the authenticated user.

Parameters:

Parameter	Type	Required	Description
limit	number	No	Maximum results (default: 10)
auth_token	string	No	Firebase ID token

Example:

{
  "limit": 5
}

Response: Array of pre-mortem summaries with IDs, names, status, and dates.

---

premortem.queue

Queue an asynchronous pre-mortem job for longer/complex analyses.

Parameters:

Parameter	Type	Required	Description
input	object	Yes	Same as premortem.run
auth_token	string	No	Firebase ID token

Response: Job ID for tracking progress.

---

premortem.status

Check the status of a queued pre-mortem job.

Parameters:

Parameter	Type	Required	Description
jobId	string	Yes	Job ID from premortem.queue
auth_token	string	No	Firebase ID token

Response: Job status, progress percentage, and current stage.

---

premortem.kick

Execute or resume a queued pre-mortem job.

Parameters:

Parameter	Type	Required	Description
jobId	string	Yes	Job ID to execute
auth_token	string	No	Firebase ID token

---

premortem.regen_assumptions

Regenerate the assumptions section of an existing pre-mortem.

Parameters:

Parameter	Type	Required	Description
input	object	Yes	Original input parameters
doc	object	Yes	Existing pre-mortem document
auth_token	string	No	Firebase ID token

---

premortem.regen_experiments

Regenerate the experiments section of an existing pre-mortem.

Parameters:

Parameter	Type	Required	Description
input	object	Yes	Original input parameters
doc	object	Yes	Existing pre-mortem document
auth_token	string	No	Firebase ID token

---

Persona Tools

These tools are available through the cutline-tools server.

personas.list

List all personas for the authenticated user.

Parameters:

Parameter	Type	Required	Description
productId	string	No	Filter by product ID
auth_token	string	Yes	Firebase ID token

Response: Array of persona objects with demographics, personality, and behaviors.

---

personas.get

Get a specific persona by ID.

Parameters:

Parameter	Type	Required	Description
personaId	string	Yes	Persona ID
auth_token	string	Yes	Firebase ID token

Response: Complete persona object.

---

personas.chat

Have a conversation with a persona about your product or feature.

Parameters:

Parameter	Type	Required	Description
persona	object	Yes	Persona object
userMessage	string	Yes	Your message to the persona
conversationHistory	array	No	Previous messages for context
product	object	No	Product context
auth_token	string	Yes	Firebase ID token

Example:

{
  "persona": {
    "name": "Debra",
    "description": "Technical solopreneur running multiple SaaS businesses",
    "role": "Founder",
    "segment": "Startup"
  },
  "userMessage": "What do you think about adding a mobile app to our analytics platform?",
  "product": {
    "name": "Cutline",
    "brief": "AI product management suite"
  }
}

Response: Persona's response in character, based on their profile.

---

Wiki Tools

These tools are available through the cutline-tools server.

wiki.load

Load the wiki/documentation for a project.

Parameters:

Parameter	Type	Required	Description
projectId	string	Yes	Project ID
auth_token	string	No	Firebase ID token

Response: Wiki content in markdown format.

---

wiki.save

Save updated wiki content.

Parameters:

Parameter	Type	Required	Description
projectId	string	Yes	Project ID
markdown	string	Yes	Wiki content in markdown
auth_token	string	No	Firebase ID token

---

wiki.apply_edits

Apply specific edits to wiki content.

Parameters:

Parameter	Type	Required	Description
edits	array	Yes	Array of edit operations
auth_token	string	No	Firebase ID token

---

Agent Tools

agent.chat

Chat with the Cutline AI assistant about your wiki or product.

Parameters:

Parameter	Type	Required	Description
prompt	string	Yes	Your question or request
wikiMarkdown	string	No	Wiki context for the chat
auth_token	string	No	Firebase ID token

Example:

{
  "prompt": "What are the main risks identified in my product wiki?",
  "wikiMarkdown": "# My Product\n\n## Risks\n- Market timing..."
}

---

trial.generate

Generate a trial run scenario for testing.

Parameters:

Parameter	Type	Required	Description
prompt	string	Yes	Description of what to test

Example:

{
  "prompt": "Generate a trial run for testing our checkout flow with 100 users"
}

---

Output Tools

These tools are available through the cutline-output server.

premortem.render_pdf

Generate a PDF from a pre-mortem document.

Parameters:

Parameter	Type	Required	Description
doc	object	Yes	Pre-mortem document
store	boolean	No	Store PDF in cloud (default: false)
auth_token	string	No	Firebase ID token

Response: Signed URL to download the PDF.

---

premortem.qa_answer

Answer questions about a pre-mortem document.

Parameters:

Parameter	Type	Required	Description
doc	object	Yes	Pre-mortem document
question	string	Yes	Question to answer

Example:

{
  "doc": { ... },
  "question": "What are the top 3 risks identified?"
}

---

Integration Tools

These tools are available through the cutline-integrations server.

integrations.create_issues

Create issues in an external issue tracker from pre-mortem results.

Parameters:

Parameter	Type	Required	Description
schema_json	object	Yes	Issue schema and data
limit	number	No	Maximum issues to create
auth_token	string	No	Firebase ID token

---

Error Handling

All tools may return errors in this format:

{
  "error": "MCP error -32600: Permission denied"
}

Common Error Codes:

Code	Meaning	Solution
-32600	Permission denied	Check authentication
-32601	Method not found	Verify tool name
-32602	Invalid params	Check required parameters
-32603	Internal error	Contact support

Rate Limits

Tier	Engineering Audits	Code Audits	Pre-mortems
Free	3/month	—	—
Premium	Unlimited	Unlimited	Unlimited
Enterprise	Unlimited	Unlimited	Unlimited

Next Steps

• Setup Guide - Get started with MCP
• Examples - Common workflows