@evalops/maestro 0.10.18 → 0.10.20

package/skills/incident-triage/SKILL.md

@@ -0,0 +1,40 @@
+---
+name: incident-triage
+description: Triage production incidents by building a timeline, scoping blast radius, identifying mitigations, and preserving evidence. Use when the user asks about an outage, regression, alert, Sentry issue, or incident follow-up.
+license: Complete terms in LICENSE.txt
+compatibility: "Maestro skill packages with bounded GitHub MCP tools and executable toolbox entries."
+allowed-tools:
+  - read
+  - search
+  - bash
+builtin-tools:
+  - read
+  - search
+mode: incident
+isolatedContext: true
+metadata:
+  version: "0.1.0"
+  category: evalops-operations
+  artifactSchema: evalops.maestro.skill.incident_triage.v1
+---
+
+# Incident Triage
+
+Use this skill to turn scattered incident evidence into a concise operator plan.
+
+## Workflow
+
+1. Confirm the symptom, start time, affected surface, and urgency.
+2. Load `reference/triage.md` for the detailed timeline and mitigation checklist.
+3. Gather only scoped evidence from granted tools and local files. Keep customer data and internal handles out of the normal answer.
+4. Build a timeline with known, inferred, and unknown entries separated.
+5. Identify the likely owner, immediate mitigation, verification signal, and follow-up issue.
+6. End with current state, blast radius, next action, and what evidence was withheld or unavailable.
+
+## Toolbox
+
+Run `toolbox/incident-timeline` to emit the required incident report skeleton.
+
+## MCP Scope
+
+The bundled GitHub MCP config exposes issue, code, workflow, and file lookups for incident context. Live observability systems must be granted by the runtime separately.

package/skills/incident-triage/mcp.json

@@ -0,0 +1,14 @@
+{
+  "github": {
+    "command": "npx",
+    "args": ["-y", "@modelcontextprotocol/server-github"],
+    "includeTools": [
+      "search_issues",
+      "get_issue",
+      "list_issue_comments",
+      "search_code",
+      "get_file_contents",
+      "list_workflow_runs"
+    ]
+  }
+}

package/skills/incident-triage/reference/triage.md

@@ -0,0 +1,20 @@
+# Incident Triage Reference
+
+## Timeline Discipline
+
+- Known: directly observed logs, alerts, commits, deploys, or operator reports.
+- Inferred: plausible links that need confirmation.
+- Unknown: missing evidence that affects the decision.
+
+## Mitigation Checklist
+
+1. Is the incident still active?
+2. What changed before the first symptom?
+3. Which customers, tenants, or environments are affected?
+4. Is there a reversible mitigation?
+5. What signal proves recovery?
+6. What follow-up issue or post-incident artifact is needed?
+
+## Answer Shape
+
+Use concise sections: current state, blast radius, likely cause, mitigation, verification, next action, withheld or unavailable evidence.

package/skills/incident-triage/toolbox/incident-timeline

@@ -0,0 +1,22 @@
+#!/usr/bin/env bash
+set -euo pipefail
+
+if [ "${MAESTRO_TOOLBOX_ACTION:-}" = "describe" ]; then
+  cat <<'JSON'
+{"name":"incident-timeline","description":"Prints the required incident triage timeline and mitigation sections.","inputs":["incident","start_time","surface"],"outputs":["timeline","blast_radius","mitigation","next_action"]}
+JSON
+  exit 0
+fi
+
+cat <<'TEXT'
+Current state:
+Blast radius:
+Timeline:
+- Known:
+- Inferred:
+- Unknown:
+Mitigation:
+Verification:
+Next action:
+Withheld or unavailable evidence:
+TEXT

package/skills/incident-triage/toolbox/incident-timeline.cmd

@@ -0,0 +1,16 @@
+@echo off
+if /I "%MAESTRO_TOOLBOX_ACTION%"=="describe" (
+  echo {"name":"incident-timeline","description":"Prints the required incident triage timeline and mitigation sections.","inputs":["incident","start_time","surface"],"outputs":["timeline","blast_radius","mitigation","next_action"]}
+  exit /b 0
+)
+
+echo Current state:
+echo Blast radius:
+echo Timeline:
+echo - Known:
+echo - Inferred:
+echo - Unknown:
+echo Mitigation:
+echo Verification:
+echo Next action:
+echo Withheld or unavailable evidence:

package/skills/pr-review/SKILL.md

@@ -0,0 +1,40 @@
+---
+name: pr-review
+description: Review GitHub pull requests for correctness, regressions, missing tests, and merge readiness. Use when the user asks for PR review, review feedback, or a pre-merge risk pass.
+license: Complete terms in LICENSE.txt
+compatibility: "Maestro skill packages with bounded GitHub MCP tools and executable toolbox entries."
+allowed-tools:
+  - read
+  - search
+  - bash
+builtin-tools:
+  - read
+  - search
+mode: review
+isolatedContext: true
+metadata:
+  version: "0.1.0"
+  category: evalops-operations
+  artifactSchema: evalops.maestro.skill.pr_review.v1
+---
+
+# PR Review
+
+Use this skill to produce a merge-readiness review, not a broad design memo.
+
+## Workflow
+
+1. Identify the PR, base branch, changed files, linked issue, CI status, and review history.
+2. Load `reference/rubric.md` only when you need the detailed severity rubric or checklist.
+3. Inspect the changed code and nearby tests. Prefer local repo reads for code and bounded GitHub MCP calls for PR metadata.
+4. Lead with findings ordered by severity. Include file and line when the issue is in code.
+5. Call out missing tests only when they hide real risk.
+6. If no issues are found, say that clearly and name the remaining residual risk.
+
+## Toolbox
+
+Run `toolbox/review-summary` when you need the expected output sections without loading the reference file.
+
+## MCP Scope
+
+The bundled GitHub MCP config exposes only pull-request, review, file, and check metadata needed for review. Do not ask it for broad organization scans from this skill.

package/skills/pr-review/mcp.json

@@ -0,0 +1,14 @@
+{
+  "github": {
+    "command": "npx",
+    "args": ["-y", "@modelcontextprotocol/server-github"],
+    "includeTools": [
+      "get_pull_request",
+      "list_pull_request_files",
+      "list_pull_request_reviews",
+      "list_pull_request_comments",
+      "get_file_contents",
+      "list_check_runs"
+    ]
+  }
+}

package/skills/pr-review/reference/rubric.md

@@ -0,0 +1,20 @@
+# PR Review Rubric
+
+## Severity
+
+- Critical: data loss, auth bypass, secret exposure, production outage, or a change that cannot be safely rolled back.
+- High: user-visible regression, broken deploy path, incorrect persisted state, race condition, or unhandled error path.
+- Medium: missing validation, incomplete migration, flaky workflow, or test gap on changed behavior.
+- Low: maintainability issue that is likely to slow future work but does not block merge.
+
+## Review Checklist
+
+1. Confirm the PR changes the behavior it claims to change.
+2. Trace changed inputs through persistence, queues, caches, and external APIs.
+3. Check rollback and partial-failure behavior.
+4. Verify tests exercise the risky branch, not only the happy path.
+5. Check docs or operator runbooks when the change affects release, incident, or customer support behavior.
+
+## Output Contract
+
+Write findings first. Each finding should include impact, evidence, and a concrete fix direction. Keep summaries short and put them after findings.

package/skills/pr-review/toolbox/review-summary

@@ -0,0 +1,20 @@
+#!/usr/bin/env bash
+set -euo pipefail
+
+if [ "${MAESTRO_TOOLBOX_ACTION:-}" = "describe" ]; then
+  cat <<'JSON'
+{"name":"review-summary","description":"Prints the required PR review output sections for Maestro first-party review skills.","inputs":["pr","repo","risk_level"],"outputs":["findings","open_questions","verification"]}
+JSON
+  exit 0
+fi
+
+cat <<'TEXT'
+Findings:
+- [severity] file:line - impact and concrete fix direction
+
+Open questions:
+- None.
+
+Verification:
+- State the exact checks inspected or run.
+TEXT

package/skills/pr-review/toolbox/review-summary.cmd

@@ -0,0 +1,14 @@
+@echo off
+if /I "%MAESTRO_TOOLBOX_ACTION%"=="describe" (
+  echo {"name":"review-summary","description":"Prints the required PR review output sections for Maestro first-party review skills.","inputs":["pr","repo","risk_level"],"outputs":["findings","open_questions","verification"]}
+  exit /b 0
+)
+
+echo Findings:
+echo - [severity] file:line - impact and concrete fix direction
+echo.
+echo Open questions:
+echo - None.
+echo.
+echo Verification:
+echo - State the exact checks inspected or run.

package/skills/release-verification/SKILL.md

@@ -0,0 +1,40 @@
+---
+name: release-verification
+description: Verify release readiness and post-release health from PRs, CI, tags, deploy evidence, rollback notes, and operator artifacts. Use when the user asks whether a release can ship, has shipped, or is safe to promote.
+license: Complete terms in LICENSE.txt
+compatibility: "Maestro skill packages with bounded GitHub MCP tools and executable toolbox entries."
+allowed-tools:
+  - read
+  - search
+  - bash
+builtin-tools:
+  - read
+  - search
+mode: ship
+isolatedContext: true
+metadata:
+  version: "0.1.0"
+  category: evalops-operations
+  artifactSchema: evalops.maestro.skill.release_verification.v1
+---
+
+# Release Verification
+
+Use this skill to prove release state from artifacts, not from branch vibes.
+
+## Workflow
+
+1. Identify the release candidate, target environment, owning repo, and rollback boundary.
+2. Load `reference/checklist.md` when the release has deploy, migration, or customer-visible risk.
+3. Inspect CI, release notes, tags, deployment references, and outstanding blockers.
+4. Separate required evidence from optional confidence checks.
+5. Report allowed evidence links or revisions, unavailable sources, blockers, next action, and what was withheld or out of scope.
+6. Never call a release shipped until the artifact or deployment state supports it.
+
+## Toolbox
+
+Run `toolbox/release-readiness` to emit the required release-verification report skeleton.
+
+## MCP Scope
+
+The bundled GitHub MCP config is limited to release, workflow, pull request, and check metadata. Use platform or deploy-specific tools only when the runtime grants them separately.

package/skills/release-verification/mcp.json

@@ -0,0 +1,15 @@
+{
+  "github": {
+    "command": "npx",
+    "args": ["-y", "@modelcontextprotocol/server-github"],
+    "includeTools": [
+      "get_pull_request",
+      "list_check_runs",
+      "list_workflow_runs",
+      "get_workflow_run",
+      "list_workflow_jobs",
+      "list_releases",
+      "get_release_by_tag"
+    ]
+  }
+}

package/skills/release-verification/reference/checklist.md

@@ -0,0 +1,21 @@
+# Release Verification Checklist
+
+## Required Evidence
+
+- Candidate revision, tag, or build receipt.
+- Required CI checks and their final states.
+- Deployment target and environment ownership.
+- Migration, feature flag, or config changes.
+- Rollback path and rollback owner.
+- Known customer or operator impact.
+
+## Report Shape
+
+Use this order:
+
+1. Decision: ready, blocked, already promoted, or needs more evidence.
+2. Evidence: links, revisions, check names, and artifact IDs that are safe to cite.
+3. Unavailable sources: what could not be verified and why.
+4. Blockers: only blockers that change the release decision.
+5. Next action: the smallest action that moves the release forward.
+6. Withheld or out of scope: customer data, internal handles, or evidence that should stay in durable runtime state.

package/skills/release-verification/toolbox/release-readiness

@@ -0,0 +1,18 @@
+#!/usr/bin/env bash
+set -euo pipefail
+
+if [ "${MAESTRO_TOOLBOX_ACTION:-}" = "describe" ]; then
+  cat <<'JSON'
+{"name":"release-readiness","description":"Prints the required release verification report skeleton.","inputs":["repo","revision","environment"],"outputs":["decision","evidence","blockers","next_action"]}
+JSON
+  exit 0
+fi
+
+cat <<'TEXT'
+Decision:
+Evidence:
+Unavailable sources:
+Blockers:
+Next action:
+Withheld or out of scope:
+TEXT

package/skills/release-verification/toolbox/release-readiness.cmd

@@ -0,0 +1,12 @@
+@echo off
+if /I "%MAESTRO_TOOLBOX_ACTION%"=="describe" (
+  echo {"name":"release-readiness","description":"Prints the required release verification report skeleton.","inputs":["repo","revision","environment"],"outputs":["decision","evidence","blockers","next_action"]}
+  exit /b 0
+)
+
+echo Decision:
+echo Evidence:
+echo Unavailable sources:
+echo Blockers:
+echo Next action:
+echo Withheld or out of scope: