Skip to content

Configure — Customize the Session

The configure tool manages your Gasoline session — filter noise, store data, manage recordings, compare error states, diff sessions, control streaming, and view audit logs.

Need one runnable call + response shape + failure fix for every mode? See Configure Executable Examples.

configure({what:"noise_rule", noise_action: "auto_detect"}) // Auto-filter noise
configure({what:"store", store_action: "save", key: "baseline", data: {...}}) // Save data
configure({what:"clear", buffer: "all"}) // Clear all buffers
configure({what:"health"}) // Server health
configure({what:"recording_start"}) // Start recording
configure({what:"recording_stop", recording_id: "rec-123"}) // Stop recording
configure({what:"playback", recording_id: "rec-123"}) // Replay recording
configure({what:"log_diff", original_id: "rec-1", replay_id: "rec-2"}) // Compare error states
configure({what:"diff_sessions", verif_session_action: "capture", name: "v1"}) // Session snapshot
configure({what:"audit_log", tool_name: "observe"}) // View tool usage history

These parameters are shared across many configure modes:

ParameterTypeDescription
whatstring (required)Mode to execute (noise_rule, store, diff_sessions, etc.)
actionstringDeprecated alias for what
telemetry_modestringTelemetry metadata mode: off, auto, full
tab_idnumberOptional target tab ID for tab-aware modes
operationstringSecondary operation selector used by mode-specific flows
limitnumberMax entries returned for list/report modes

Manages noise rules that suppress irrelevant errors — browser extension noise, analytics failures, framework internals.

configure({what:"noise_rule", noise_action: "auto_detect"})

Scans current errors and identifies patterns that are likely noise (extension errors, analytics, third-party scripts). Creates rules automatically.

configure({what:"noise_rule",
noise_action: "add",
pattern: "analytics\\.google",
category: "console",
reason: "Google Analytics noise"})
ParameterTypeDescription
noise_actionstringadd, remove, list, reset, auto_detect
patternstringRegex pattern to match against error messages
categorystringBuffer: console, network, or websocket
reasonstringHuman-readable explanation of why this is noise
rulesarrayBatch-add multiple rules at once
rule_idstringID of rule to remove (for remove action)
configure({what:"noise_rule", noise_action: "list"})
configure({what:"noise_rule", noise_action: "remove", rule_id: "rule-123"})
configure({what:"noise_rule", noise_action: "reset"})

Save and load JSON data that persists across sessions. Useful for storing baselines, configuration, or any data the AI needs to remember.

configure({what:"store",
store_action: "save",
namespace: "baselines",
key: "homepage-vitals",
data: {lcp: 1200, cls: 0.05}})
configure({what:"store",
store_action: "load",
namespace: "baselines",
key: "homepage-vitals"})
configure({what:"store", store_action: "list"})
configure({what:"store", store_action: "list", namespace: "baselines"})
configure({what:"store", store_action: "delete", key: "homepage-vitals"})
configure({what:"store", store_action: "stats"})
ParameterTypeDescription
store_actionstringsave, load, list, delete, stats
namespacestringLogical grouping for keys
keystringStorage key
dataobjectJSON data to persist (for save)

Load the saved session context snapshot (project ID, baseline list, noise config, schema hints, and performance summary) from the server-side session store.

configure({what: "load"})

Remove captured data from memory.

configure({what:"clear", buffer: "all"})
configure({what:"clear", buffer: "network"})
configure({what:"clear", buffer: "logs"})
ParameterTypeDescription
bufferstringnetwork, websocket, actions, logs, all

Start capturing a browser session. Records user actions and browser state for later playback or comparison.

configure({what:"event_recording_start"})
configure({what:"event_recording_start", name: "checkout-run"})

Stop an active recording session.

configure({what:"event_recording_stop", recording_id: "rec-123"})
ParameterTypeDescription
recording_idstringID of the recording to stop

Replay a previously captured recording.

configure({what:"playback", recording_id: "rec-123"})
ParameterTypeDescription
recording_idstringID of the recording to replay

Compare error states between two recordings. Useful for verifying that a bug fix resolved the issue, or detecting regressions after a deploy.

configure({what:"log_diff", original_id: "rec-abc", replay_id: "rec-xyz"})
ParameterTypeDescription
original_idstringOriginal recording ID (baseline)
replay_idstringReplay recording ID (comparison)

telemetry — Configure Telemetry Metadata

Section titled “telemetry — Configure Telemetry Metadata”

Set the global telemetry metadata mode. Individual tool calls can override this with the telemetry_mode parameter.

configure({what:"telemetry", telemetry_mode: "auto"})
configure({what:"telemetry", telemetry_mode: "off"})
ParameterTypeDescription
telemetry_modestringoff, auto, or full

Enable or disable real-time event notifications. When enabled, Gasoline proactively notifies the AI about errors, performance regressions, and security issues as they happen.

configure({what:"streaming", streaming_action: "enable",
events: ["errors", "performance", "security"],
severity_min: "warning",
throttle_seconds: 5})
configure({what:"streaming", streaming_action: "disable"})
configure({what:"streaming", streaming_action: "status"})
ParameterTypeDescription
streaming_actionstringenable, disable, status
eventsarrayCategories: errors, network_errors, performance, user_frustration, security, regression, anomaly, ci, all
severity_minstringMinimum severity: info, warning, error
throttle_secondsintegerMinimum seconds between notifications (1-60)

Check the Gasoline server’s status, uptime, buffer usage, and connected clients.

configure({what:"health"})

No additional parameters. Returns server version, uptime, buffer occupancy, client count, and rate limit status.


test_boundary_start / test_boundary_end — Test Boundaries

Section titled “test_boundary_start / test_boundary_end — Test Boundaries”

Mark the start and end of a test run. Events within boundaries can be correlated for test-specific analysis.

configure({what:"test_boundary_start", test_id: "checkout-flow", label: "Guest Checkout Test"})
// ... run the test ...
configure({what:"test_boundary_end", test_id: "checkout-flow"})
ParameterTypeDescription
test_idstringUnique identifier for the test
labelstringHuman-readable description

diff_sessions — Session Snapshots & Comparison

Section titled “diff_sessions — Session Snapshots & Comparison”

Capture named snapshots of the current session state and compare them to detect changes.

configure({what:"diff_sessions", verif_session_action: "capture", name: "before-deploy"})
configure({what:"diff_sessions",
verif_session_action: "compare",
compare_a: "before-deploy",
compare_b: "after-deploy"})
configure({what:"diff_sessions", verif_session_action: "list"})
configure({what:"diff_sessions", verif_session_action: "delete", name: "old-snapshot"})
ParameterTypeDescription
verif_session_actionstringcapture, compare, list, delete
namestringSnapshot name
compare_astringFirst snapshot (baseline)
compare_bstringSecond snapshot (comparison target)

View a log of all MCP tool calls made during the session. Useful for debugging, compliance, and understanding AI agent behavior.

configure({what:"audit_log"})
configure({what:"audit_log", tool_name: "observe", limit: 20})
configure({what:"audit_log", since: "2026-02-07T10:00:00Z"})
ParameterTypeDescription
tool_namestringFilter by tool name
audit_session_idstringFilter by audit session ID
sincestringOnly entries after this ISO 8601 timestamp
limitnumberMaximum entries to return

describe_capabilities — Tool Capability Discovery

Section titled “describe_capabilities — Tool Capability Discovery”

Returns a description of the tool’s capabilities. Useful for AI agents to understand what actions are available.

configure({what: "describe_capabilities"})
configure({what: "describe_capabilities", tool: "observe", mode: "errors"})
ParameterTypeDescription
toolstringFilter to a specific tool (e.g., observe, interact)
modestringFilter to a specific mode within that tool

Force-restart the Gasoline daemon when it becomes unresponsive. Works even when the daemon is completely hung.

configure({what: "restart"})

Run a diagnostic health check. Verifies binary, port, extension connection, and client configuration.

configure({what: "doctor"})

security_mode — Debug Mode for Altered Environments

Section titled “security_mode — Debug Mode for Altered Environments”

Opt into an altered-environment debug mode for advanced troubleshooting scenarios like proxied traffic inspection.

configure({what: "security_mode"}) // Read current mode
configure({what: "security_mode", mode: "insecure_proxy", confirm: true}) // Enable
configure({what: "security_mode", mode: "normal"}) // Disable
ParameterTypeDescription
modestringnormal or insecure_proxy
confirmbooleanRequired true when enabling insecure_proxy

Save and replay named sequences of interact actions. Useful for repeatable workflows — demo scripts, test setup flows, or multi-step automation.

configure({what: "save_sequence",
name: "login-flow",
description: "Log in as test user",
steps: [
{what: "navigate", url: "https://example.com/login"},
{what: "type", selector: "label=Email", text: "test@example.com"},
{what: "click", selector: "text=Sign In"}
],
tags: ["auth", "setup"]})
ParameterTypeDescription
namestring (required)Sequence name
stepsarray (required)Ordered list of interact actions
descriptionstringHuman-readable description
tagsarrayLabels for categorization
configure({what: "replay_sequence", name: "login-flow"})
configure({what: "replay_sequence", name: "login-flow", step_timeout_ms: 15000, stop_after_step: 2})
ParameterTypeDescription
namestring (required)Sequence name to replay
override_stepsarraySparse array of step overrides (null = use saved)
step_timeout_msnumberTimeout per step (default 10000)
continue_on_errorbooleanContinue if a step fails (default true)
stop_after_stepnumberStop after executing this many steps

get_sequence / list_sequences / delete_sequence

Section titled “get_sequence / list_sequences / delete_sequence”
configure({what: "get_sequence", name: "login-flow"})
configure({what: "list_sequences"})
configure({what: "delete_sequence", name: "login-flow"})

tutorial / examples — Quick Start Guidance

Section titled “tutorial / examples — Quick Start Guidance”

Return quickstart snippets and context-aware setup guidance.

configure({what: "tutorial"})
configure({what: "examples"})

network_recording — Network Traffic Recording

Section titled “network_recording — Network Traffic Recording”

Manage network traffic recording for specific domains.

configure({what: "network_recording", operation: "start", domain: "api.example.com"})
configure({what: "network_recording", operation: "status"})
configure({what: "network_recording", operation: "stop"})
ParameterTypeDescription
operationstringstart, stop, or status
domainstringDomain to record traffic for

Configure random delays before interact actions. Useful for making automated flows look more natural or for stress-testing race conditions.

configure({what: "action_jitter", action_jitter_ms: 500})
configure({what: "action_jitter", action_jitter_ms: 0}) // Disable
ParameterTypeDescription
action_jitter_msnumberMaximum random delay in milliseconds (0 to disable)

Create or preview a structured issue report directly from the running session. This bundles environment context, recent diagnostics, and your notes.

configure({what: "report_issue", operation: "list_templates"})
configure({what: "report_issue", operation: "preview", template: "bug", user_context: "Daemon disconnected while replaying"})
configure({what: "report_issue", operation: "submit", template: "bug", title: "Replay disconnects intermittently", user_context: "Happens after ~20 actions"})
ParameterTypeDescription
operationstringlist_templates, preview, or submit
templatestringIssue template name
titlestringIssue title (required for submit)
user_contextstringYour repro notes/context attached to the report