Claude code plugins with skills for agentic development
postindustria-tech/agentic-toolkit
Agent
Researches how a task should be implemented in THIS codebase for the dev-practices execute skill's research atom. Surfaces the existing primitives/components to reuse, the CURRENT (not deprecated) pattern to follow, and the Core Invariant -- recording findings into the beads task. Spawned by the execute skill in a…
postindustria-tech/agentic-toolkit
Agent
Independent adversarial reviewer for the dev-practices execute skill's review atom. Reads a beads task's research findings and implementation plan, rates findings LOW/MEDIUM/HIGH against a fixed rubric, and returns a verdict (ALLLOW / NEEDSREFINEMENT / NEEDSUSERINPUT) that routes the next atom. Spawned by the execute…
postindustria-tech/agentic-toolkit
Agent
Writes the failing regression test (TDD red) for the dev-practices execute skill's write-test atom. Designs the test at the right level -- integration/ E2E through the outer surface by default -- from the plan stored in a beads task. Spawned by the execute skill in a fresh context -- NOT for direct or proactive…
postindustria-tech/agentic-toolkit
Agent
Reviews naming conventions, error message formats, API response shapes, logging patterns, and config access for cross-module consistency. Read-only.
postindustria-tech/agentic-toolkit
Agent
Single-disease, whole-tree codebase scanner for the in-loop targeted review (code-review --targeted). Given ONE disease-pattern spec and ONE lens (dedup | consistency | layering), finds every instance of that ONE pattern across the whole source tree and emits a disposition table (MIGRATE / ALLOWLIST / DEFER).…
postindustria-tech/agentic-toolkit
Agent
Reviews code for logic duplication -- not just textual copy-paste but semantically equivalent code expressed differently. Especially important in AI-assisted codebases where generated code varies syntactically. Read-only.
postindustria-tech/agentic-toolkit
Agent
Reviews whether logic lives in the correct architectural layer: transport wrappers vs business logic vs repositories vs adapters vs services. Read-only.
postindustria-tech/agentic-toolkit
Agent
Reviews code for Pythonic idioms, Pydantic best practices, SQLAlchemy 2.0 patterns, and async correctness. Read-only -- writes findings to output file.
postindustria-tech/agentic-toolkit
Agent
Reviews test quality: are tests testing behavior or just mocking everything? Are assertions meaningful? Is coverage real or illusory? Do mocks verify the right things? Read-only -- writes findings to output file.
postindustria-tech/agentic-toolkit
Agent
Use this agent when LangGraph workflow files are created or modified to validate structure and detect anti-patterns. Examples.