@hoststack.dev/mcp 0.6.0 → 0.7.0

package/README.md

@@ -79,22 +79,24 @@ If `HOSTSTACK_API_KEY` is set in your shell, it gets baked into the snippet; oth
 
 ## Tool inventory
 
-45 tools, grouped by resource:
-
-| Category         | Read                                                                      | Write                                                                          |
-| ---------------- | ------------------------------------------------------------------------- | ------------------------------------------------------------------------------ |
-| **projects**     | `list_projects`, `get_project`                                            | `create_project`, `update_project`                                             |
-| **services**     | `list_services`, `get_service`, `get_service_metrics`, `get_service_logs` | `update_service`, `update_service_config`, `suspend_service`, `resume_service` |
-| **deploys**      | `list_deploys`, `get_deploy`, `get_deploy_logs`                           | `trigger_deploy`, `cancel_deploy`                                              |
-| **environments** | `list_environments`                                                       | `create_environment`, `delete_environment`, `promote_deploy`                   |
-| **databases**    | `list_databases`, `get_database`                                          | — (use the dashboard for create/delete/credentials)                            |
-| **volumes**      | `list_volumes`                                                            | `create_volume`, `update_volume`, `delete_volume`                              |
-| **domains**      | `list_domains`                                                            | `add_domain`, `verify_domain`, `remove_domain`                                 |
-| **dns**          | `list_dns_zones`, `list_dns_records`, `get_dns_record`                    | `create_dns_record`, `update_dns_record`, `delete_dns_record`                  |
-| **env-vars**     | `list_env_vars`                                                           | `set_env_var`, `delete_env_var`, `bulk_set_env_vars`                           |
-| **cron**         | `list_cron_executions`, `get_cron_execution`                              | —                                                                              |
-| **activity-log** | `list_activity_log`                                                       | —                                                                              |
-| **meta**         | `get_me`                                                                  | —                                                                              |
+55 tools, grouped by resource:
+
+| Category          | Read                                                                                                                              | Write                                                                                                                    |
+| ----------------- | --------------------------------------------------------------------------------------------------------------------------------- | ------------------------------------------------------------------------------------------------------------------------ |
+| **projects**      | `list_projects`, `get_project`                                                                                                    | `create_project`, `update_project`                                                                                       |
+| **services**      | `list_services`, `get_service`, `get_service_metrics`, `get_service_metrics_history`, `get_service_logs`, `get_service_logs_bulk` | `update_service`, `update_service_config`, `suspend_service`, `resume_service`                                           |
+| **deploys**       | `list_deploys`, `get_deploy`, `get_deploy_logs`, `diagnose_deploy`                                                                | `trigger_deploy`, `cancel_deploy`                                                                                        |
+| **environments**  | `list_environments`                                                                                                               | `create_environment`, `delete_environment`, `promote_deploy`                                                             |
+| **databases**     | `list_databases`, `get_database`                                                                                                  | `update_database` (use the dashboard for create/delete/credentials)                                                      |
+| **volumes**       | `list_volumes`                                                                                                                    | `create_volume`, `update_volume`, `delete_volume`                                                                        |
+| **domains**       | `list_domains`                                                                                                                    | `add_domain`, `verify_domain`, `remove_domain`                                                                           |
+| **dns**           | `list_dns_zones`, `list_dns_records`, `get_dns_record`                                                                            | `create_dns_record`, `update_dns_record`, `delete_dns_record`                                                            |
+| **env-vars**      | `list_env_vars`                                                                                                                   | `set_env_var`, `delete_env_var`, `bulk_set_env_vars`                                                                     |
+| **cron**          | `list_cron_executions`, `get_cron_execution`                                                                                      | —                                                                                                                        |
+| **notifications** | `list_notification_channels`                                                                                                      | `create_notification_channel`, `update_notification_channel`, `delete_notification_channel`, `test_notification_channel` |
+| **alerts**        | `list_alerts`                                                                                                                     | —                                                                                                                        |
+| **activity-log**  | `list_activity_log`                                                                                                               | —                                                                                                                        |
+| **meta**          | `get_me`                                                                                                                          | —                                                                                                                        |
 
 A few design notes worth knowing as a caller:
 

package/dist/hoststack-mcp.js

@@ -483,21 +483,31 @@ defineTool({
     "",
     'By default events are AGGREGATED by (action, resourceId) so flapping events collapse to one row with a fire count + first/last timestamps \u2014 e.g. "service.auto_restarted on service 31, 8 times in the last hour, last at 14:22". Pass aggregate=false to see every raw row.',
     "",
+    'ACTIVE vs HISTORICAL (v89): each aggregated entry now has an `active` boolean and a `lastResolvedAt` timestamp. `active=true` means the most recent occurrence has NOT been resolved \u2014 the alert is still on fire. The default `active=true` filter hides resolved/historical alerts so triage starts with "what is broken right now". Pass `active=false` to include resolved entries too (useful for post-mortems). Resolution is automatic for some alert kinds \u2014 e.g. `deploy.failed_consecutive` is cleared when the next deploy of the same service goes live.',
+    "",
+    'Database backup_failed entries (v89) include `containerStatus` / `containerHealth` / `containerExitCode` in lastMetadata when the agent could resolve the container at failure time, so you can tell "Redis is overloaded" apart from "Redis is restarting" without a second tool call.',
+    "",
+    "Deploy.failed_consecutive entries (v89) carry the offending `commitHash` in lastMetadata, and the streak now dedupes per (service, commitHash) \u2014 one critical per bad commit, not one every 6 retries.",
+    "",
     "Inputs (all optional):",
     '  - since: ISO-8601 timestamp OR relative offset like "-1h" / "-2d". Default: -24h.',
     "  - until: ISO-8601 upper bound (ignored when aggregating \u2014 aggregated view always extends to now).",
     "  - limit: max rows (default 100, hard cap 500).",
     "  - aggregate: true (default) collapses by (action, resourceId); false returns raw rows.",
+    "  - active: true (default) shows only alerts that are still on fire (resolved_at IS NULL on the most recent row). false includes resolved historical alerts.",
     "",
-    "Returns: { alerts: Array, aggregated: boolean }. Each aggregated entry includes action, resourceType, resourceId, severity, count, firstFiredAt, lastFiredAt, lastMetadata. Each raw entry includes id, action, resourceType, resourceId, severity, metadata, createdAt.",
+    "Returns: { alerts: Array, aggregated: boolean, activeOnly: boolean }. Each aggregated entry includes action, resourceType, resourceId, severity, count, firstFiredAt, lastFiredAt, lastResolvedAt (nullable), active (boolean), lastMetadata. Each raw entry adds resolvedAt (nullable).",
     "",
-    "Example: list_alerts({ since: '-1h' }) \u2192 { alerts: [{ action: 'deploy.failed', resourceId: 31, severity: 'error', count: 3, lastFiredAt: '\u2026', \u2026 }] }."
+    "Example: list_alerts({ since: '-1h' }) \u2192 { alerts: [{ action: 'deploy.failed_consecutive', resourceId: 31, severity: 'critical', active: true, count: 3, lastFiredAt: '\u2026', lastResolvedAt: null, lastMetadata: { commitHash: 'abc1234' }, \u2026 }] }."
   ].join("\n"),
   input: {
     since: z3.string().optional().describe('ISO-8601 timestamp or relative offset (e.g. "-1h", "-2d"). Default: -24h.'),
     until: z3.string().optional().describe("ISO-8601 upper bound. Only honored when aggregate=false."),
     limit: z3.number().int().positive().max(500).optional().describe("Max rows (default 100, hard cap 500)."),
-    aggregate: z3.boolean().optional().describe("Collapse by (action, resourceId). Default true.")
+    aggregate: z3.boolean().optional().describe("Collapse by (action, resourceId). Default true."),
+    active: z3.boolean().optional().describe(
+      "Only return alerts still on fire (resolved_at IS NULL on the most recent row). Default true. Pass false to include resolved historical alerts."
+    )
   },
   handler: async (args2, ctx) => {
     const teamId = await ctx.resolveTeamId();
@@ -506,11 +516,13 @@ defineTool({
     if (args2.until !== void 0) params["until"] = args2.until;
     if (args2.limit !== void 0) params["limit"] = String(args2.limit);
     if (args2.aggregate === false) params["aggregate"] = "0";
+    if (args2.active === false) params["active"] = "0";
     const response = await ctx.api.get(`/api/alerts/${teamId}`, params);
     const items = Array.isArray(response.alerts) ? response.alerts.map(shape) : [];
     const aggregated = Boolean(response.aggregated);
-    const summary = items.length === 0 ? "No alerts in the requested window \u2014 everything is operating normally." : aggregated ? `Returned ${items.length} alert group${items.length === 1 ? "" : "s"} (flapping events collapsed).` : `Returned ${items.length} raw alert event${items.length === 1 ? "" : "s"}.`;
-    return respond({ summary, data: { alerts: items, aggregated } });
+    const activeOnly = args2.active !== false;
+    const summary = items.length === 0 ? activeOnly ? "No active alerts in the requested window \u2014 everything is operating normally." : "No alerts in the requested window (active or resolved)." : aggregated ? `Returned ${items.length} alert group${items.length === 1 ? "" : "s"}${activeOnly ? " (active only \u2014 pass active=false for resolved history)" : ""} (flapping events collapsed).` : `Returned ${items.length} raw alert event${items.length === 1 ? "" : "s"}.`;
+    return respond({ summary, data: { alerts: items, aggregated, activeOnly } });
   }
 });
 
@@ -613,7 +625,7 @@ defineTool({
     "Inputs:",
     '  - database_id: publicId of the database (e.g. "db_\u2026").',
     "  - name (optional): new database name.",
-    '  - plan (optional): "free" | "starter" | "standard" | "pro" \u2014 changes memory tier.',
+    '  - plan (optional): "free" | "micro" | "starter" | "standard" | "pro" \u2014 changes memory tier.',
     "  - disk_size_gb (optional): new disk size in GB (must be \u2265 current).",
     "",
     "Returns: { database: Database } \u2014 the updated record.",
@@ -623,7 +635,7 @@ defineTool({
   input: {
     database_id: z5.string().describe("Database publicId (e.g. db_xyz)."),
     name: z5.string().min(1).max(100).optional().describe("New database name."),
-    plan: z5.enum(["free", "starter", "standard", "pro"]).optional().describe("Plan tier (memory/CPU)."),
+    plan: z5.enum(["free", "micro", "starter", "standard", "pro"]).optional().describe("Plan tier (memory/CPU)."),
     disk_size_gb: z5.number().int().min(1).max(1024).optional().describe("New disk size in GB. Must be \u2265 current.")
   },
   handler: async (args2, ctx) => {
@@ -682,7 +694,7 @@ defineTool({
     "Inputs:",
     '  - service_id: publicId of the service (e.g. "svc_abc123"). Use list_services to find it.',
     "",
-    "Returns: { items: Deploy[] } \u2014 each deploy includes id, publicId, status (pending|building|deploying|live|failed|cancelled), commitSha, commitMessage, branch, triggeredBy, startedAt, finishedAt, buildDurationMs (just the docker build / image-pull step), totalDurationMs (full deploy wall-clock = finishedAt \u2212 startedAt).",
+    'Returns: { items: Deploy[] } \u2014 each deploy includes id, publicId, status (pending|building|deploying|live|failed|cancelled), commitSha, commitMessage, branch, triggeredBy, startedAt, finishedAt, imageBuildMs (docker build / image-pull only; null on skip-build redeploys \u2014 v89), containerBootMs (deploying \u2192 live wall-clock; null on builds that failed before container start \u2014 v89), buildDurationMs (legacy alias of imageBuildMs kept for back-compat), totalDurationMs (full deploy wall-clock = finishedAt \u2212 startedAt). Use imageBuildMs + containerBootMs together to tell "build is slow" apart from "boot is slow".',
     "",
     'Example: list_deploys({ service_id: "svc_abc" }) \u2192 { items: [{ publicId: "dpl_\u2026", status: "live", commitMessage: "Fix login", \u2026 }, \u2026] }'
   ].join("\n"),
@@ -709,7 +721,7 @@ defineTool({
     "  - service_id: publicId of the service.",
     '  - deploy_id: publicId of the deploy (e.g. "dpl_\u2026").',
     "",
-    "Returns: { deploy: Deploy } \u2014 full deploy record (status, commitSha, commitMessage, branch, buildDurationMs, totalDurationMs, finishedAt, etc).",
+    "Returns: { deploy: Deploy } \u2014 full deploy record (status, commitSha, commitMessage, branch, imageBuildMs + containerBootMs (v89 split timings; legacy buildDurationMs preserved as alias), totalDurationMs, finishedAt, etc).",
     "",
     'Example: get_deploy({ service_id: "svc_abc", deploy_id: "dpl_xyz" }) \u2192 { deploy: { status: "live", \u2026 } }'
   ].join("\n"),
@@ -1703,7 +1715,9 @@ defineTool({
     type: z11.enum(["slack", "discord", "email"]).describe("Channel type."),
     name: z11.string().min(1).max(128).describe("Human-readable label."),
     webhook_url: z11.string().max(500).describe("Slack/Discord webhook URL or email address (when type=email)."),
-    events: z11.array(z11.enum(NOTIFICATION_EVENTS)).describe("List of events the channel subscribes to. Empty list = subscribe to nothing.")
+    events: z11.array(z11.enum(NOTIFICATION_EVENTS)).describe(
+      "List of events the channel subscribes to. Empty list = subscribe to nothing."
+    )
   },
   handler: async (args2, ctx) => {
     const teamId = await ctx.resolveTeamId();
@@ -2004,24 +2018,30 @@ defineTool({
   name: "get_service",
   category: "services",
   description: [
-    "Fetch a single service by ID, including its current status and configuration summary.",
+    "Fetch a single service by ID with its current status AND its service_config row (resources, health-check tuning, scaling, restart policy).",
     "",
-    "When to use: drilling into a service after list_services, checking deploy/runtime status, or grabbing the repo+branch before triggering a deploy.",
+    "When to use: drilling into a service after list_services, checking deploy/runtime status, grabbing the repo+branch before triggering a deploy, or inspecting the health-check / autoscale knobs before tweaking them via update_service_config.",
     "",
     "Inputs:",
     '  - service_id: publicId of the service (e.g. "svc_abc123").',
     "",
-    "Returns: { service: Service } \u2014 type, status, runtime, repoUrl, branch, autoDeploy, region, plan, createdAt, updatedAt.",
+    "Returns: { service: Service, config: ServiceConfig } \u2014 service has type/status/runtime/repoUrl/branch/autoDeploy/region/plan/timestamps; config has memoryMb, cpuShares, diskSizeGb, port, protocol, healthCheckEnabled, healthCheckInterval, healthCheckTimeout, healthCheckGracePeriodSec, restartPolicy, preDeployCommand, min/maxInstances, scale thresholds.",
     "",
-    'Example: get_service({ service_id: "svc_abc" }) \u2192 { service: { type: "web", status: "running", \u2026 } }'
+    'Example: get_service({ service_id: "svc_abc" }) \u2192 { service: { type: "web", status: "running", \u2026 }, config: { healthCheckGracePeriodSec: 120, \u2026 } }'
   ].join("\n"),
   input: {
     service_id: z13.string().describe("Service publicId (e.g. svc_abc123).")
   },
   handler: async (args2, ctx) => {
     const teamId = await ctx.resolveTeamId();
-    const response = await ctx.hoststack.services.get(teamId, args2.service_id);
-    const data = { service: shapeService(response.service) };
+    const [serviceResponse, configResponse] = await Promise.all([
+      ctx.hoststack.services.get(teamId, args2.service_id),
+      ctx.hoststack.services.getConfig(teamId, args2.service_id)
+    ]);
+    const data = {
+      service: shapeService(serviceResponse.service),
+      config: shape(configResponse.config)
+    };
     const status = data.service && "status" in data.service ? data.service.status : "unknown";
     return respond({ summary: `Service ${args2.service_id} is ${status}.`, data });
   }
@@ -2133,9 +2153,9 @@ defineTool({
   name: "update_service_config",
   category: "services",
   description: [
-    "Update build/runtime configuration for a service: build command, start command, install command, branch, root directory, dockerfile path, auto-deploy flag, instance count. All fields optional \u2014 pass only what you want to change.",
+    "Update build/runtime configuration for a service. All fields optional \u2014 pass only what you want to change.",
     "",
-    "When to use: the user wants to tweak how a service builds or runs. Build/runtime fields (branch, install/build/start command, root, dockerfile) take effect on the next deploy \u2014 call trigger_deploy after if you need them applied immediately. instance_count rescales without a redeploy.",
+    "When to use: the user wants to tweak how a service builds, runs, scales, or health-checks. Build/runtime fields (branch, install/build/start command, root, dockerfile) take effect on the next deploy \u2014 call trigger_deploy after if you need them applied immediately. Instance_count, resource and health-check changes rescale or rewire without a redeploy.",
     "",
     "Inputs:",
     "  - service_id: publicId of the service.",
@@ -2144,12 +2164,26 @@ defineTool({
     "  - root_directory (optional): build context root inside the repo.",
     "  - dockerfile_path (optional): path to Dockerfile relative to root_directory. Pass null to clear.",
     "  - auto_deploy (optional): boolean \u2014 auto-deploy on git push.",
-    "  - instance_count (optional): integer \u22651 \u2014 pin both min and max instances to this value.",
+    '  - health_check_path (optional): HTTP path the platform GETs to verify liveness (e.g. "/health"). Pass null for TCP-only check.',
+    "  - health_check_enabled (optional): boolean \u2014 toggle health checking on/off.",
+    "  - health_check_interval (optional): integer 5\u2013300 seconds \u2014 how often the check runs.",
+    "  - health_check_timeout (optional): integer 1\u201360 seconds \u2014 single-attempt timeout.",
+    '  - health_check_grace_period_sec (optional): integer 1\u20131800 seconds \u2014 startup tolerance before failures count. RAISE THIS (e.g. 180) when the agent reports "Health check timed out" on a cold-boot app (Bun + Vite SSR typically need 90\u2013180s).',
+    "  - memory_mb (optional): integer 128\u201316384 \u2014 container memory cap.",
+    "  - cpu_shares (optional): integer 128\u20134096 \u2014 relative CPU weight.",
+    "  - disk_size_gb (optional): integer 1\u2013100 \u2014 ephemeral disk cap.",
+    "  - port (optional): integer 1\u201365535 \u2014 container port the platform forwards traffic to.",
+    '  - protocol (optional): "http" | "tcp".',
+    '  - restart_policy (optional): "always" | "on-failure" | "no".',
+    "  - pre_deploy_command (optional): shell command run before the new release accepts traffic (typical use: migrations).",
+    "  - instance_count (optional): integer 1\u201350 \u2014 pin both min and max instances to this value.",
+    "  - min_instances, max_instances (optional): integers \u2014 autoscale bounds. Use instead of instance_count when you want a range.",
+    "  - scale_cpu_threshold, scale_memory_threshold (optional): integer 10\u2013100 \u2014 autoscale trigger percentage.",
     '  - log_filter_rules (optional): list of { pattern, action } rules applied to runtime logs at query time. Pattern matches the message by case-insensitive substring; action is "drop" (filter out) or "downgrade" (flip stderr \u2192 stdout so it stops looking like an error). Pass [] to clear all rules. Capped at 50 rules.',
     "",
     "Returns: { service?: Service, config?: ServiceConfig } \u2014 whichever rows were touched.",
     "",
-    'Example: update_service_config({ service_id: "svc_abc", start_command: "bun apps/api/src/index.ts" }) \u2192 { service: { startCommand: "bun apps/api/src/index.ts", \u2026 } }'
+    'Example: update_service_config({ service_id: "svc_abc", health_check_grace_period_sec: 180 }) \u2192 { config: { healthCheckGracePeriodSec: 180, \u2026 } }'
   ].join("\n"),
   input: {
     service_id: z13.string().describe("Service publicId."),
@@ -2160,7 +2194,25 @@ defineTool({
     root_directory: z13.string().optional().describe("Build context root."),
     dockerfile_path: z13.string().nullable().optional().describe("Path to Dockerfile relative to root. Null clears."),
     auto_deploy: z13.boolean().optional().describe("Auto-deploy on push."),
+    health_check_path: z13.string().nullable().optional().describe('HTTP health-check path (e.g. "/health"). Null = TCP-only check.'),
+    health_check_enabled: z13.boolean().optional().describe("Toggle health checking on/off."),
+    health_check_interval: z13.number().int().min(5).max(300).optional().describe("How often the check runs, in seconds (5\u2013300)."),
+    health_check_timeout: z13.number().int().min(1).max(60).optional().describe("Single-attempt timeout in seconds (1\u201360)."),
+    health_check_grace_period_sec: z13.number().int().min(1).max(1800).optional().describe(
+      "Startup grace period in seconds (1\u20131800). Raise this if the app needs more time to boot before health checks start counting failures."
+    ),
+    memory_mb: z13.number().int().min(128).max(16384).optional().describe("Container memory cap in MB (128\u201316384)."),
+    cpu_shares: z13.number().int().min(128).max(4096).optional().describe("Relative CPU weight (128\u20134096)."),
+    disk_size_gb: z13.number().int().min(1).max(100).optional().describe("Ephemeral disk size in GB (1\u2013100)."),
+    port: z13.number().int().min(1).max(65535).optional().describe("Container port the platform forwards traffic to."),
+    protocol: z13.enum(["http", "tcp"]).optional().describe("Traffic protocol."),
+    restart_policy: z13.enum(["always", "on-failure", "no"]).optional().describe("Docker restart policy."),
+    pre_deploy_command: z13.string().optional().describe("Shell command run before the new release accepts traffic."),
     instance_count: z13.number().int().positive().max(50).optional().describe("Pin min and max instances to this value (1\u201350)."),
+    min_instances: z13.number().int().min(0).max(50).optional().describe("Autoscale lower bound. Use with max_instances for a range."),
+    max_instances: z13.number().int().min(1).max(50).optional().describe("Autoscale upper bound. Use with min_instances for a range."),
+    scale_cpu_threshold: z13.number().int().min(10).max(100).optional().describe("Autoscale CPU trigger percentage (10\u2013100)."),
+    scale_memory_threshold: z13.number().int().min(10).max(100).optional().describe("Autoscale memory trigger percentage (10\u2013100)."),
     log_filter_rules: z13.array(
       z13.object({
         pattern: z13.string().min(1).max(200),
@@ -2182,11 +2234,35 @@ defineTool({
     if (args2.branch !== void 0) serviceUpdate["branch"] = args2.branch;
     if (args2.root_directory !== void 0) serviceUpdate["rootDirectory"] = args2.root_directory;
     if (args2.auto_deploy !== void 0) serviceUpdate["autoDeploy"] = args2.auto_deploy;
+    if (args2.health_check_path !== void 0)
+      serviceUpdate["healthCheckPath"] = args2.health_check_path;
     const configUpdate = {};
+    if (args2.health_check_enabled !== void 0)
+      configUpdate["healthCheckEnabled"] = args2.health_check_enabled;
+    if (args2.health_check_interval !== void 0)
+      configUpdate["healthCheckInterval"] = args2.health_check_interval;
+    if (args2.health_check_timeout !== void 0)
+      configUpdate["healthCheckTimeout"] = args2.health_check_timeout;
+    if (args2.health_check_grace_period_sec !== void 0)
+      configUpdate["healthCheckGracePeriodSec"] = args2.health_check_grace_period_sec;
+    if (args2.memory_mb !== void 0) configUpdate["memoryMb"] = args2.memory_mb;
+    if (args2.cpu_shares !== void 0) configUpdate["cpuShares"] = args2.cpu_shares;
+    if (args2.disk_size_gb !== void 0) configUpdate["diskSizeGb"] = args2.disk_size_gb;
+    if (args2.port !== void 0) configUpdate["port"] = args2.port;
+    if (args2.protocol !== void 0) configUpdate["protocol"] = args2.protocol;
+    if (args2.restart_policy !== void 0) configUpdate["restartPolicy"] = args2.restart_policy;
+    if (args2.pre_deploy_command !== void 0)
+      configUpdate["preDeployCommand"] = args2.pre_deploy_command;
     if (args2.instance_count !== void 0) {
       configUpdate["minInstances"] = args2.instance_count;
       configUpdate["maxInstances"] = args2.instance_count;
     }
+    if (args2.min_instances !== void 0) configUpdate["minInstances"] = args2.min_instances;
+    if (args2.max_instances !== void 0) configUpdate["maxInstances"] = args2.max_instances;
+    if (args2.scale_cpu_threshold !== void 0)
+      configUpdate["scaleCpuThreshold"] = args2.scale_cpu_threshold;
+    if (args2.scale_memory_threshold !== void 0)
+      configUpdate["scaleMemoryThreshold"] = args2.scale_memory_threshold;
     if (args2.log_filter_rules !== void 0) {
       configUpdate["logFilterRules"] = args2.log_filter_rules;
     }