@noresponse/mcp 0.4.11 → 0.5.0
This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
- package/CHANGELOG.md +35 -0
- package/README.md +5 -1
- package/bin/mcp-server.js +925 -39
- package/bin/mcp-server.js.map +35 -9
- package/manifest.json +17 -1
- package/package.json +1 -1
package/manifest.json
CHANGED
|
@@ -40,7 +40,7 @@
|
|
|
40
40
|
"type": "string"
|
|
41
41
|
}
|
|
42
42
|
},
|
|
43
|
-
"version": "0.
|
|
43
|
+
"version": "0.5.0",
|
|
44
44
|
"tools": [
|
|
45
45
|
{
|
|
46
46
|
"name": "create_assessment",
|
|
@@ -82,6 +82,14 @@
|
|
|
82
82
|
"name": "record_enforcement",
|
|
83
83
|
"description": "Record what you did to an account in your identity provider.\n\nRecords what you did to the account in your identity provider for one flow's decision at one risk_version: the action (or none), the account's state before it, the provider and whether it was applied or failed. It is your action, not an outcome, and it leaves the account's risk unchanged. A record is kept once per account, risk_version, flow and action: the first report answers 201, and a repeat answers 200 with the stored record unchanged, so retries are safe without an Idempotency-Key. An applied report after a failed one replaces the failed record, keeping its id and created_at, and answers 201. A risk_version older than the account's current one is accepted and stored, but only an applied record at the current risk_version takes the account out of listChanged with unenforced=true; a newer one answers 409 risk_version_ahead. An account with no current risk answers 404 not_found. error_code is allowed only with status failed. Requires the history scope."
|
|
84
84
|
},
|
|
85
|
+
{
|
|
86
|
+
"name": "list_keys",
|
|
87
|
+
"description": "List API keys.\n\nLists the keys of the signing key's environment, newest first, at most 200, revoked and expired keys included, without secrets: each key's id, prefix, scopes, status, and who created and revoked it. Requires the management scope. Read-only and safe to retry."
|
|
88
|
+
},
|
|
89
|
+
{
|
|
90
|
+
"name": "revoke_key",
|
|
91
|
+
"description": "Revoke an API key.\n\nRevokes a key of the signing key's environment at once: its next request answers 401. Requires the management scope. A key of the other environment or of another workspace answers 404 not_found. A key cannot revoke itself (409 self_revocation): rotate it with keys.rotate, or revoke it with another management key. Revoking a revoked key changes nothing and answers the key as it was revoked, so a retry is safe."
|
|
92
|
+
},
|
|
85
93
|
{
|
|
86
94
|
"name": "create_integration",
|
|
87
95
|
"description": "Register a hosted provider integration.\n\nRegisters a WorkOS or Auth0 integration and returns its hosted endpoint, url, so a sign-up or sign-in is scored with no server of your own. For WorkOS, send the Actions signing secret from the WorkOS dashboard and point the user registration and authentication Actions at url. For Auth0, the answer's secret is the shared secret of the NoResponse Auth0 Action, shown once: store it as an Action secret. hook_points defaults to both pre_registration and post_login; an event that is not listed is allowed without an assessment. The endpoint waits for the full assessment for at most 2.5 seconds (WorkOS) or 4.5 seconds (Auth0), denies only when risk.block is true and allows on any failure. Requires the integrations scope; at most 10 per account. Not idempotent: a retry after a lost answer registers another integration."
|
|
@@ -126,6 +134,14 @@
|
|
|
126
134
|
"name": "replay_webhook_delivery",
|
|
127
135
|
"description": "Send one delivery again.\n\nQueues the event for the endpoint again: the same event id and body it received, with a new NR-Webhook-Timestamp and a signature from the endpoint's current signing key, retried like any other delivery and listed with replay: true. Your receiver deduplicates on the event id. The endpoint must be verified. Requires the webhooks scope. Not idempotent: each call, a retry included, sends the event again."
|
|
128
136
|
},
|
|
137
|
+
{
|
|
138
|
+
"name": "get_settings",
|
|
139
|
+
"description": "Read an environment's settings.\n\nReads the settings of the signing key's environment: retention in days; per flow (account_creation, login, purchase) the step_up, review and block thresholds, the default depth and the admission; and the sources the environment refuses. A flow without thresholds of its own shows account_creation's with inherits_account_creation true. Requires the management scope; the other environment answers 404 not_found. Read-only and safe to retry."
|
|
140
|
+
},
|
|
141
|
+
{
|
|
142
|
+
"name": "update_settings",
|
|
143
|
+
"description": "Change an environment's settings.\n\nChanges only what the body names, in one transaction: retention_days, refused_sources, and per flow step_up, review, block (null turns blocking off), default_depth, admission, hold_seconds, on_expiry, or inherits_account_creation to return login or purchase to account_creation's thresholds. After the change each flow keeps 0 < step_up < review <= 100 and block off or review <= block <= 100; a change that breaks it answers 400 naming the flow. A policy change is one new policy_version with the signing key's prefix as updated_by, and every change is recorded in the workspace's audit log with the key as its actor. Requires the management scope; the other environment answers 404 not_found. A change reaches assessments within a minute."
|
|
144
|
+
},
|
|
129
145
|
{
|
|
130
146
|
"name": "wait_for_assessment",
|
|
131
147
|
"description": "Wait until an assessment is complete, then return it.\n\nPolls get_assessment at the interval the API's Retry-After asks for (at least once a second) until status is complete, or until timeout_seconds pass. When the time runs out the result says so and carries the latest answer, and you can call this tool again. A partial answer has a recommendation you can act on while a lane is still running; a pending one has none, so apply your own fallback until it settles. Requires the assess or history scope."
|