Give Claude Code the
context it deserves.
Claude Code is powerful — but it only knows what you paste in. Pathmode's MCP server feeds it structured intent specs, dependency graphs, and real user evidence. Every implementation stays grounded in reality.
> Using MCP tool: get_current_intent
> Fetching active intent from Pathmode...
objective: "Redesign account settings so users can update preferences without drop-off"
outcomes: 3 testable conditions
constraints: 3 non-negotiable rules
evidence: 5 items from user research
> Ready. Starting Phase 1: Orient...
The context gap is real
AI agents ship fast. But without structured context, they ship guesses.
Without Pathmode
- ×Requirements scattered across Notion, Jira, Slack, and your head
- ×Agent sees a prompt — not the user pain behind it
- ×No dependency awareness — new code breaks old flows
- ×Verification is a manual vibe check
- ×"Done" means it compiled, not that it solved the problem
With Pathmode
- Every spec traces to real user friction, quotes, and metrics
- Agent gets objective, outcomes, constraints, and edge cases
- Dependency graph prevents breaking upstream flows
- Phase-gated execution: orient, plan, implement, validate
- Agent marks intent verified — the loop closes
The closed-loop system
Evidence flows in. Code flows out. Verification loops back.
Evidence Board
Friction, quotes, metrics from real users
Intent Spec
Objective, outcomes, constraints, edge cases
Claude Code
Orient → Plan → Implement → Validate
Verified
Status updated, decisions logged back
12 tools. Full context.
Claude Code connects via Model Context Protocol and gets instant access to your entire Intent Layer.
Read Intent
get_current_intentActive intent with full spec
get_intentSingle intent by ID
list_intentsAll intents, filter by status
search_intentsKeyword search across specs
Understand Strategy
analyze_intent_graphCritical path, bottlenecks, cycles
get_intent_relationsDependency graph for one intent
get_workspaceVision, principles, non-negotiables
get_constitutionMandatory constraint rules
Execute & Track
get_agent_promptStructured execution prompt
export_contextCLAUDE.md, .cursorrules, intent.md
update_intent_statusMark shipped or verified
log_implementation_noteRecord technical decisions
What Claude Code sees when it calls get_current_intent
{
"objective": "Redesign account settings so users can update preferences without drop-off",
"severity": "critical",
"outcomes": [
"Settings save confirmation visible within 500ms",
"Password change completes in under 3 clicks",
"Drop-off rate on settings page below 30%"
],
"constraints": [
"Must work on iOS Safari 15+",
"No full-page reloads on save",
"Respect existing RBAC permissions"
],
"edgeCases": [
{ "scenario": "User loses connection mid-save", "expected": "Draft persisted locally, retry on reconnect" },
{ "scenario": "Concurrent edits from two sessions", "expected": "Last-write-wins with conflict toast" }
],
"evidence": "5 items — 3 friction reports, 1 user quote, 1 drop-off metric"
}Phase-gated execution
Pathmode doesn't just hand over a spec. It structures how Claude Code works through it.
Orient
Read relevant source files. Understand existing patterns and conventions. Identify conflicts between spec and codebase.
Gate — confirm understanding before proceeding
Plan
List every file to change. Map changes to outcomes. Order by dependencies. Every outcome in the spec must be covered.
Implement
Execute in dependency order. Respect all constraints — constitutional rules are non-negotiable. Pause and reassess on unexpected behavior.
Validate
Run every verification step from the spec — E2E tests, unit tests, manual checks. If anything fails, fix and re-run. Mark intent as shipped.
Not just fast. Correct.
Most agents race straight to implementation. They hallucinate forward — writing code before understanding context, skipping verification because nothing told them not to.
Pathmode's phase-gated prompts force Claude Code to orient first, plan second, implement third, and validate last. The gates aren't suggestions — they're checkpoints baked into the execution prompt.
Agent confirms understanding before writing a single line
Every outcome in the spec maps to a planned change
Constitutional constraints can't be skipped or forgotten
Verification steps are mandatory, not optional
Setup in under 2 minutes
One config file. One API key. That's it.
Get your API key
Go to Settings → API Keys in your Pathmode workspace and generate a key.
Add to your Claude Code config
Paste this into .claude/settings.json
{
"mcpServers": {
"pathmode": {
"command": "npx",
"args": ["@pathmode/mcp-server"],
"env": {
"PATHMODE_API_KEY": "pm_live_..."
}
}
}
}Start coding
Claude Code now has full access to your Intent Layer. Ask it to get_current_intent and start building.
Works offline too — add --local to read intent.md files from your project directory.
Stop pasting specs.
Start shipping intents.
Give Claude Code the context it deserves.
No credit card required · Free for solo builders