agents-def

Plugin agents-def from marketplace kolezka in kolezka/marketplace. Commands read from the repo's manifest — 2 days ago.

Skills (1)

• orchestrator

  Orchestrate complex engineering tasks by delegating exploration, implementation, testing, and review to specialized subagents instead of doing all work in the main context.

Subagents (9)

• architect

  Use this agent when system boundaries, contracts, ownership, or migration choices are unresolved. Typical triggers include comparing durable designs, planning a backwards-compatible migration, and resolving responsibility across components. Do not use it for a routine implementation plan with a sett

• codebase-researcher

  Use this agent when a task needs a coherent subsystem model across modules. Typical triggers include tracing state and data flow, integration boundaries, persistence, and background work. Do not use it for a quick symbol search or to implement a change.

• debugger

  Use this agent when observed behavior is wrong and the root cause is unclear. Typical triggers include a failing test, an unexpected production symptom, or fixes that did not work. Do not use it for a known implementation task or a broad architecture decision.

• explorer

  Use this agent for fast repository location work. Typical triggers include finding files, symbols, call sites, established patterns, and related tests before a known change. Do not use it to model a subsystem across modules or design a solution.

• implementer

  Use this agent to implement a known, well-scoped solution in an assigned worktree. Typical triggers include an approved fix, a bounded feature, or a mechanical refactor with clear acceptance criteria. Do not use it to settle architecture or investigate an unclear failure.

• reviewer

  Use this agent for an independent review of an explicit diff or named change set. Typical triggers include checking correctness, regressions, compatibility, and maintainability after implementation. Do not use it as proof that an unexamined system is correct or for a security-focused review.

• security-reviewer

  Use this agent for a defensive security review of an explicit code or change scope. Typical triggers include assessing trust boundaries, authentication and authorization, untrusted inputs, secrets, and dependencies. Do not use it as a general review or for active attacks and scanning.

• test-engineer

  Use this agent to design or implement focused tests for specified behavior. Typical triggers include adding regression coverage, defining observable acceptance cases, and testing important edge paths. Do not use it to repair production code outside the brief.

• verifier

  Use this agent to independently check stated acceptance criteria after a change. Typical triggers include validating existing tests or builds, exercising an isolated fixture, and confirming a scenario in an approved environment. Do not use it to write code or tests.

Similar plugins