atlassian-orchestratorRouter for the Acendas Atlassian Suite. Use when the task spans multiple products (Jira + Confluence + Bitbucket + QMetry), when the right specialist is unclear, for any QMetry test-management request (no specialist exists - handled here via the as-qmetry-* skills), or when the user asks to fetch/sh
code-review-orchestratorAutonomous multi-step code review of Bitbucket pull requests with Jira context. Owns the parallel scanner pipeline (security, bugs, silent failures, patterns, tests, spec, plus conditional specialists), the deep-dive investigators, and an adversarial critic. Triggers: 'review PR #X', 'deep review of
devops-orchestratorAutonomous Bitbucket DevOps work - Pipelines runs and logs, deployments, environments, branch protection, code insights, and pipeline schedules/variables. Triggers: 'audit CI failures last week', 'why are pipelines failing', 'promote to staging', 'audit branch protection on main', 'what is deployed
knowledge-orchestratorAutonomous Confluence knowledge-base work - reading, writing, editing, and restructuring pages and spaces. Triggers: 'edit confluence page X', 'audit our runbook tree', 'restructure docs in space ENG', 'create a multi-page docs hierarchy', 'diff confluence page versions', 'publish this content to co
pr-criticAdversarial PR critic with anti-sycophancy directive. Challenges PRs against their Jira spec using assumption extraction, pre-mortem narrative, and structured criteria. Surfaces blind spots the scanners miss. Read-only. Spawned by code-review-orchestrator on high-stakes PRs (large diff, payments/aut
pr-investigatorDeep-dive investigator for a single high-stakes PR review finding. Receives ONE finding from a scanner and confirms or refutes it with evidence — reads call sites, traces data flow, checks tests, examines git blame on the destination branch. Returns a verdict (CONFIRMED / REFUTED / NEEDS-USER) with
pr-review-bugsPR logic-bug scanner. Single responsibility — finds correctness errors only (off-by-one, null/undefined handling, race conditions, resource leaks, wrong operators, type confusion). Spawned by code-review-orchestrator with a target PR + file list. Read-only. Returns structured findings with confidenc
pr-review-contractsPR API contract & backward compatibility scanner. Single responsibility — finds breaking changes to public interfaces, request/response shape drift, schema breaks, deprecation gaps, and consumer impact. Spawned by code-review-orchestrator. Read-only. Confidence ≥ 80.
pr-review-databasePR database scanner. Single responsibility — finds problematic queries, migrations, indexing issues, locking risks, and data-consistency problems. Spawned by code-review-orchestrator on diffs touching db/migration/query files. Read-only. Confidence ≥ 80.
pr-review-financialPR financial-app scanner. Single responsibility — finds money/ledger/auth/audit/idempotency/reconciliation risks specific to financial systems. Spawned conditionally by code-review-orchestrator when the diff touches payment/billing/ledger/wallet code OR the project marks itself as financial. Read-on
pr-review-observabilityPR observability scanner. Single responsibility — finds gaps in logs, metrics, traces, and error visibility that will make production debugging hard. Spawned by code-review-orchestrator. Read-only. Confidence ≥ 80.
pr-review-patternsPR conventions/patterns scanner. Single responsibility — finds violations of project rules, naming conventions, anti-patterns, duplication, dead code, magic numbers. Reads project rules from .claude/rules/ if present. Spawned by code-review-orchestrator. Read-only. Confidence ≥ 80.
pr-review-performancePR performance & scalability scanner. Single responsibility — finds hot-path regressions, unnecessary allocations, complexity blowups, async/sync mistakes, resource leaks, scalability issues. Spawned by code-review-orchestrator. Read-only. Confidence ≥ 80.
pr-review-rolloutPR rollout & operational safety scanner. Single responsibility — finds risky configuration defaults, missing feature flags on risky changes, dependency churn, and rollout/recovery gaps. Spawned by code-review-orchestrator. Read-only. Confidence ≥ 80.
pr-review-securityPR security scanner. Single responsibility — finds security vulnerabilities only (injection, auth/authz bypass, hardcoded secrets, crypto misuse, unsafe deserialization, input validation gaps, path traversal, SSRF). Spawned by code-review-orchestrator. Read-only. Returns findings with confidence ≥ 8
pr-review-silent-failuresPR silent-failure scanner. Single responsibility — finds code that fails quietly (empty catches, swallowed errors, masked failures, retries that hide root cause). The most underreported bug class. Spawned by code-review-orchestrator. Read-only. Confidence ≥ 80.
pr-review-specPR spec compliance scanner. Single responsibility — maps acceptance criteria from the linked Jira issue(s) to the PR's code, flags missing implementations and over-building (functionality beyond spec). Spawned by code-review-orchestrator. Read-only. Confidence ≥ 80.
pr-review-testsPR test quality scanner. Single responsibility — finds missing critical-path coverage, weak assertions, missing edge cases, brittle tests, missing error-path tests. Cross-references test files with implementation. Spawned by code-review-orchestrator. Read-only. Confidence ≥ 80.
release-orchestratorAutonomous release management - drafting release notes from merged PRs and closed Jira issues, managing fixVersions, tagging, and publishing release docs to Confluence. Triggers: 'draft release notes for v1.4', 'what shipped this week', 'tag and publish v2.0', 'set fixVersion across these issues', '
sprint-orchestratorAutonomous Jira Agile sprint and backlog work - backlog ordering, planning from backlog, retro briefs, standup briefs, and active-sprint health management. Triggers: 'reorder the backlog', 'move ABC-12 above ABC-7', 'rank these to the top', 'plan next sprint from backlog', 'sprint retro for the plat
triage-orchestratorAutonomous Jira issue triage - bulk categorization, label/component/priority suggestions, dedupe detection, linking issues to PRs and Confluence docs, batch transitions, and structured epic/sub-task creation. Triggers: 'triage incoming bugs', 'categorize unprocessed issues', 'find duplicates in this