Overview

Collections group one or more HTTP API endpoints together so they can be run as a single regression suite. Each run compares the live response schema against a saved baseline and surfaces breaking changes, schema drift, and latency regressions — in the browser or from CI via the CLI.

Collections list view showing collection name, endpoint count, last run time, duration, and percentage of endpoints that passed, with Edit, Delete, and Run action buttons

The Collections view lists each collection with its endpoint count, last run stats (time, duration, pass %), and action buttons.

Creating a Collection

  1. Navigate to HTTP API Explorer → Collections.
  2. Click New Collection in the toolbar.
  3. Give the collection a name and optional description.
  4. Add endpoints from your saved HTTP API definitions.
  5. Click Save.

Running a Collection

Click the ▶ Run button on any collection row to execute all its endpoints in sequence.

Collection run result dialog showing a summary table with endpoint name, status code, latency, result label, and schema diff badges

After each run, MCP Explorer:

  • Persists aggregate stats — last run time, total duration, passed/failed counts — on the collection object, so stats survive page reloads without replaying history.
  • Persists per-endpoint summaries — status code, latency, schema comparison result, and error message for each endpoint in the run.
  • Writes a history entry — appended to HttpApiCollections/{collectionId}/history.jsonl for long-term trend analysis.

Schema Comparison Badges

Each endpoint result is decorated with one of:

BadgeMeaning
✅ PassStatus OK, schema matches baseline
⚠️ DriftSchema changed but not breaking (new optional fields)
🔴 BreakingRequired fields removed or types changed
❌ ErrorRequest failed (network error, timeout)

+ / - / ~ badges show added, removed, and changed properties respectively.

Collection Run History

Expand any collection row using the chevron toggle on the left to reveal the Run History table.

Expanded collection row showing Run History table with columns: When, Duration, Passed, Failed, Source
ColumnDescription
WhenRelative timestamp of the run
DurationTotal elapsed time for the full collection run
PassedNumber of endpoints that passed
FailedNumber of endpoints that failed or had breaking schema changes
Source⌨ CLI or 🖥 App — where the run was triggered

Expand any history row to see the same per-endpoint detail table as the run result dialog — including schema diff badges and error messages.

Tip: Run history is lazy-loaded on first expand and invalidated after each new run, so the table always reflects the latest data without an explicit page refresh.

Sparkline Trend

The Collections list also shows a Trend sparkline column for each collection — a mini bar chart of the last 10 total collection run durations. Bar colour reflects the overall health of that run (green = all passed, orange = some drift, red = failures).

Running from the CLI 

  # Run all endpoints and fail CI on breaking schema changes
mcp-http http collection run --name "Regression Suite" --fail-on-breaking

# Run against localhost when the API was saved with a Docker base URL
mcp-http http collection run --name "Regression Suite" --use-localhost --fail-on-breaking

# Run from a custom data directory
mcp-http --data-path "/path/to/data" http collection run --name "Regression Suite" --fail-on-breaking

CLI runs are recorded in history with Source = CLI, so you can distinguish automated runs from manual browser runs in the history table.

See the CLI — API & Collection Commands page for full reference.