Stateful Simulation (Alpha)

Run deterministic multi-step failure flows with session continuity.

Why this exists

Real incidents are often stateful: auth degrades over time, outage behavior ramps, and retries change outcomes. Stateless tests miss this.

When to use this vs DIY script

Use FakeMCP stateful simulation when you need profile-based transitions that are reusable in CI and demos. Use DIY scripts when you only need a one-off local state machine.

Endpoint

• POST /v1/scenario/stateful

Profiles in alpha

• token_expiry_after_n_calls
• partial_outage_ramp

Example: token expiry after N calls

export BASE_URL="https://fakemcp.com"

first="$(curl -sS -X POST "$BASE_URL/v1/scenario/stateful" \
  -H "Content-Type: application/json" \
  -d '{
    "profile": "token_expiry_after_n_calls",
    "params": { "expireAfterCalls": 2 }
  }')"

session_id="$(node -e 'const b=JSON.parse(process.argv[1]); process.stdout.write(b.sessionId)' "$first")"

curl -sS -X POST "$BASE_URL/v1/scenario/stateful" \
  -H "Content-Type: application/json" \
  -d "{
    \"profile\": \"token_expiry_after_n_calls\",
    \"sessionId\": \"${session_id}\",
    \"params\": { \"expireAfterCalls\": 2 }
  }"

Example: partial outage ramp

curl -sS -X POST "$BASE_URL/v1/scenario/stateful" \
  -H "Content-Type: application/json" \
  -d '{
    "profile": "partial_outage_ramp"
  }'

Repeat with returned sessionId to observe deterministic state transitions.