Documentation
УКР
Guide
Article Markdown

Use Case 9: Failsafe + reverse engineering

Level: basic · Modules: Scenario Builder

Required permissions and access

What you need Where in the menu Why
Scenario Builder Menu → Scenario Builder Edge audit, Runs, RE lab

If a section is missing from the menu — check your account permissions or contact ConnectiveOne support.

Business context

A production scenario must not “go silent” after an API, AI, or connect error. The integrator can read someone else’s flow, find unhandled edges, and verify the scenario before publish.

Expected result

  • In the training scenario (Use Case 1…4) every Action handles fallback / error / not_ok / limit.
  • Reverse engineering: data flow description, unhandled edges, improvement proposals.
  • Pre-publish checklist completed as self-check.

Flow architecture (audit)

For each Action node:
   Inspector → edge list → canvas → MessageKeyboard / fallback
   If an edge exists in Inspector but is not connected → scenario bug

Step-by-step implementation

Step 1. Edge audit (Use Case 1…4)

Walk scenarios Use Case 1 … Use Case 4 and for every Action node fill the table (in the results file):

Action Expected edges Connected? Leads to What the customer sees if not connected
action_fastline_pro fallback silence
action_send_request not_ok, error silence
operator_panel__connect… success, limit, error silence
registered_users__get success, error silence
custom_modules__set success, error silence

Use Case 9 rule: if an edge exists in Inspector but is not connected on the canvas — it is a scenario bug, even if it “rarely fires.”

→ Edge-driven flow

Step 2. Tooling tour — debug algorithm

Walk this sequence for any scenario problem:

Step Tool What to do
1 Runs Find the execution → stop step → which edge fired
2 Runs — trace / log Variables and constants on the problem node; error text if execution failed
3 Node Inspector → “Test action” Isolated Action test (send_request, registered_users__get, etc.)
4 Fast Line Pro → Testing Separately check categorizer / FAQ agent without a scenario
5 Instance channel Telegram / widget — when preview doesn’t reproduce keyboard, Registered Users across sessions, connect
6 Operator Line (optional E2E) After connect — is the dialog in the correct subject / skill group 
1. Runs → find the execution
2. Stop step + edge + variables/constants
3. Step log/trace (error message)
4. “Test action” in Inspector (Action) or agent test in Fast Line Pro
5. Match alias / subject_alias with Settings
6. Instance channel (Settings → Bots) — if preview is not enough
7. Operator Line — optionally after connect

→ How to test a scenario · Testing matrix

Step 3. Symptom table

Symptom Where to look What to check
Bot went silent after Action Runs trace Edge error unhandled
{{variable}} literally in text Runs → variables Variable not written
Webhook 401 / not_ok Runs → constants save_responce, Instance Settings, URL
Scenario loop Runs count Exit point
CSAT INTEGER error Keyboard source-handle Do not send free-text into CD INTEGER

Step 4. Reverse engineering lab

Complete the RE lab (~15 nodes):

  1. Describe the data flow (variables, constants).
  2. Find 3 unhandled edges.
  3. Propose improvements (fallback, validation, alias).

Or audit your own capstone scenario (Use Case 13, advanced level).

Step 5. Pre-publish checklist (self-check)

Acceptance criteria (self-check Runs)

Common mistakes

Symptom Cause What to do
“Rarely fires” — didn’t fix Edge not connected Connect all Inspector edges
Audit skipped connect Many connect blocks Same chat_closed_alias, all limit/error
RE lab — didn’t find edges Looked only at happy path Inspector of every Action
Published without Runs No trace Happy + negative path in Runs
Using your own AI agent?Install the open skill so your AI agent can work with current official ConnectiveOne documentation.Get the skillNeed support?Couldn’t find the answer or need help from the ConnectiveOne team? Create a request in Client Portal.Create a request