Team Delivery

Plugin team-delivery in Cheatr01/agents. Commands read from the repo's manifest — 1 month ago.

Skills (17)

• bug-evidence-collector

  Collect and compress source-linked logs, metrics, documentation, CI artifacts, and repository observations for a bounded defect investigation. Use for the evidence-acquisition role in BugFix; gather facts without diagnosing or changing systems.

• bug-fix

  Investigate and repair a bounded software defect with adaptive evidence collection, an explicit root-cause gate, and a handoff into Team Delivery. Use for crashes, failing tests or CI jobs, regressions, and production defects; do not use for feature discovery or open-ended incident command.

• bug-investigator

  Determine an evidence-backed root cause for a bounded software defect and produce an implementation-ready repair brief. Use after initial evidence collection in BugFix; investigate and plan, but do not implement the repair.

• contract-drift-guard

  Compare a deliberately frozen shared API, persisted schema, or cross-writer interface against implementation and decide whether a re-freeze is needed. Use only when such a contract exists.

• contract-freeze-kit

  Make a short readiness decision for a changed shared API, event, persisted schema, or interface used by parallel writers. Use before parallel implementation or a public contract change; do not use for private implementation details.

• design-freeze-kit

  Define a minimal, stable design-system decision for reusable tokens or components that multiple UI changes depend on. Use before parallel visual implementation or a design-system breaking change; do not use for a local UI tweak.

• gh-subtask-breakdown

  Create or reorganize a bounded GitHub issue hierarchy with epics, native sub-issues, milestones, labels, dependencies, and traceable branches. Use when the user asks to create, break down, or maintain GitHub tasks, epics, subtasks, or milestone work; do not use for a single-agent local change.

• independent-code-review

  Independently review every complete code diff or pull request in bounded, batched rounds, checking task fit and material business-flow effects before fixes or reports are made.

• integration-merge-governor

  Integrate and verify two or more completed parallel branches with a minimal merge order and final evidence record. Use only after multi-branch work; do not add a governance phase for a single-branch increment.

• issue-branch-guard

  Verify issue-to-branch traceability before concurrent or GitHub-governed writes. Use only when the user requires GitHub traceability or when two or more writers are active; do not block a normal single-agent local change.

• orchestration-scorekeeper

  Calculate and revalidate a compact C/R/E delivery score for a bounded software increment when a coordination, escalation, or release decision needs evidence. Do not use for trivial edits or as a reason to automatically add agents.

• performance-regression-lab

  Measure and decide on a specific observed performance regression or explicit performance target using comparable before/after evidence. Use only when a metric, workload, or regression exists; do not use for speculative tuning.

• quality-gate-matrix

  Select the smallest sufficient automated and manual validation set for a changed software increment and record a release decision. Use when regression risk is non-obvious or a release needs explicit evidence; do not use to rerun every suite after every edit.

• security-gate-runbook

  Triage material security findings and produce a compact security decision for a security-sensitive increment or release. Use when auth, permissions, secrets, sensitive data, execution, or a concrete finding is in scope; do not run a full audit by default.

• team-orchestrator

  Plan and run a bounded engineering implementation increment when the user explicitly asks for multi-agent coding coordination, role assignment, or delivery governance. Require an engineering-ready brief; do not perform product discovery, requirements analysis, or roadmap planning.

• threat-model-baseline

  Create a compact threat model for a changed trust boundary involving authentication, permissions, secrets, sensitive data, external execution, or network input. Use before implementing that boundary; do not use for unrelated local changes.

• worktree-isolation

  Set up isolated git worktrees for two or more concurrently writing subagents with already-approved branches. Use only for active parallel writes; do not create worktrees for sequential or single-agent work.

Similar plugins