Instructions file CodexOpenCode
Repository instructions for Memobank that define the project boundary, agent roles, memory rules, orchestration behavior, and development approach.
Instructions file CodexOpenCode
Repository instructions for Memobank that define the project boundary, agent roles, memory rules, orchestration behavior, and development approach.
Agent
You implement a narrowly-scoped slice of a task.
Agent
A review agent for checking whether a Memory Bank—the project's structured documentation—follows its documentation rules. It reads indexes, key documents, links, and selected project files.
Agent
You are a worker subagent. Your job is to scan a bounded part of the repo and produce a structured report.
Agent
An architecture review agent for Memory Bank documentation using the C4 model, a way to describe software from the whole system down to its parts. It checks system structure, boundaries, dependencies, documentation coverage, and key rules.
Agent
A strict code review agent that checks code quality and whether implementation matches project plans and Memory Bank documentation. It also looks for missing validation, weak error handling, mixed responsibilities, and overly complex functions.
Agent
A review assistant for implementation plans and JSON task records, which are structured files describing development work. It checks task structure, dependencies, sequencing, tests, verification steps, documentation updates, and completion requirements.
Agent
A scope and requirements review agent for Memory Bank documentation. It checks whether requirements connect to epics, features, and tasks, and whether the planned first version has clear boundaries.
Agent
A security review agent for project documentation and code. It checks threats, sensitive data, access control, common web risks, dependencies, and operational safeguards.
Agent
You do NOT implement code. You keep the protocol accurate so work can resume after context loss.
Command
A command for adding unit tests, integration tests, and end-to-end tests. These test isolated logic, connections such as databases or APIs, and complete user journeys.
Command
Optional Analysis router before PRD; recommends brainstorm, brief, write-prd, PRD decomposition, mapping, or feature clarification.
Command
An end-to-end command that takes a product requirements document (PRD) through feature and task planning, implementation, checks, and review until the work reaches a final state.
Command
An executor for an already prepared queue of JSON task records, where JSON is a structured text format for storing task details. It runs those tasks in coding sessions, usually one after another.
Command
Facilitated ideation report before Product Brief; AI coaches, user intent drives.
Command
Create or update a concise Product Brief as the PRD input contract.
Command
Optional feature-level clarification for one feature before task decomposition when needed.
Command
A command that examines a project and routes work to the appropriate starting workflow, such as analysis, a product requirements document, codebase mapping, or skeleton setup. A skeleton is the initial project structure used to hold planning information.
Command
Run a contextual interview and maintain the Project Constitution governing layer.
Command
A command for resolving unclear or conflicting requirements before coding by collecting questions, decisions, and risks.
Command
Выполнение одной TASK-XXX как implementation handoff: read -> preflight -> protocol -> implement -> local gates -> evidence -> handoff.
Command
A command that searches a project for installed skills and then checks the marketplace for missing ones.
Command
A command that examines an existing software repository and creates baseline documentation in a Memory Bank, including its architecture, components, workflows, and testing. It records what the code shows rather than planning new work.
Command
Deterministic Memory Bank readiness gate over mb-lint for autonomous/autopilot execution.