Add Agent Ops Copilot plugin preview

Author: loyal516

.github/plugin/plugin.json

@@ -0,0 +1,22 @@
+{
+  "name": "agent-ops-mini-sprint",
+  "description": "Evidence-first incident triage, PR risk review, deploy smoke, and handoff workflows for AI coding agents working on real operations issues.",
+  "version": "0.1.0",
+  "author": {
+    "name": "mysubb01",
+    "url": "https://github.com/mysubb01"
+  },
+  "repository": "https://github.com/mysubb01/agent-ops-command-pack-teaser",
+  "homepage": "https://mysubb01.github.io/agent-ops-command-pack-teaser/order.html",
+  "license": "LicenseRef-Agent-Ops-Public-Preview",
+  "keywords": [
+    "agent-ops",
+    "incident-response",
+    "code-review",
+    "deploy-smoke",
+    "handoff"
+  ],
+  "skills": [
+    "./skills/agent-ops-mini-sprint"
+  ]
+}

README.md

@@ -32,6 +32,16 @@ Fixed starter requests:
 - PR risk review: https://github.com/mysubb01/agent-ops-command-pack-teaser/issues/new?template=starter-pr-risk-review.yml
 - Deploy smoke pack: https://github.com/mysubb01/agent-ops-command-pack-teaser/issues/new?template=starter-deploy-smoke.yml
 
+## Copilot Plugin Preview
+
+This repository now includes an installable Copilot plugin preview:
+
+- Plugin name: `agent-ops-mini-sprint`
+- Manifest: [`.github/plugin/plugin.json`](./.github/plugin/plugin.json)
+- Skill: [`skills/agent-ops-mini-sprint/SKILL.md`](./skills/agent-ops-mini-sprint/SKILL.md)
+
+It packages the public Agent Ops operating workflow for incident triage, PR risk review, deploy smoke checks, and handoff notes. The plugin is free preview material; paid human-reviewed mini-sprints still start from the fast order page.
+
 ## Buy / Preview
 
 - Buy page: https://mysubb01.github.io/agent-ops-command-pack-teaser/buy.html

skills/agent-ops-mini-sprint/SKILL.md

@@ -0,0 +1,67 @@
+---
+name: agent-ops-mini-sprint
+description: 'Evidence-first operating workflow for production incident triage, PR risk review, deploy smoke checks, and handoff notes from public or sanitized context.'
+---
+
+# Agent Ops Mini-Sprint
+
+Use this skill when a user needs a small, bounded operations sprint with AI coding agents:
+
+- production bug or incident triage
+- PR risk review
+- deploy smoke checks
+- AI-generated code cleanup scoping
+- implementation handoff from public or sanitized evidence
+
+## Operating Rules
+
+- Work from evidence before naming a cause.
+- Separate observed facts, assumptions, and unknowns.
+- Use absolute timestamps when investigating incidents or deploy windows.
+- Split the path by layer: frontend, API, auth, database, queue/cache, worker/webhook, provider, runtime, and deploy.
+- Prefer read-only checks before mutations, restarts, rollbacks, or manual data edits.
+- Never request or expose secrets, tokens, private customer data, payment details, or credentials.
+- Keep the first deliverable small enough to verify in one review cycle.
+
+## Incident Triage Output
+
+Return:
+
+1. current read: what is happening, what is not proven, whether it is ongoing
+2. timeline: report time, deploys, logs, alerts, admin actions, provider events
+3. layer checks: fastest read-only proof for each plausible layer
+4. likely causes: ranked by evidence strength with why each may be wrong
+5. safe next action: smallest check or patch that increases certainty or reduces impact
+
+## PR Risk Review Output
+
+Lead with findings, not summary.
+
+For each finding include:
+
+- severity
+- file or route evidence
+- failure mode
+- missing test or verification
+- minimal fix direction
+
+Prioritize correctness, security, data loss, auth, money-path behavior, race conditions, and deploy regressions over style.
+
+## Deploy Smoke Output
+
+Return:
+
+- target environment and deploy identifier
+- health endpoints or user flows checked
+- runtime/log checks
+- rollback watchpoints
+- before/after evidence
+- handoff note for the next operator
+
+## Paid Sprint Intake
+
+If the user wants a human-reviewed mini-sprint, ask them to open a sanitized request:
+
+https://mysubb01.github.io/agent-ops-command-pack-teaser/order.html
+
+Public issues are only for sanitized context, scope, budget, and acceptance criteria. Payment links, invoices, and private handoffs happen outside public issues.