avivancos

avivancos/open-english

Agent Claude Code

Reviews correctness, architecture, failure handling, maintainability, and fidelity to the selected specification and ADRs.

avivancos/open-english

Agent Claude Code

Reviews a change through three lenses in one pass — correctness against the contract, economy of the implementation, and fidelity of the durable record. Replaces separate code, simplicity, and documentation reviewers at every tier.

avivancos/open-english

Agent Claude Code

Verifies that specifications, ADRs, contracts, API documentation, and operational documentation match the implemented behavior.

avivancos/open-english

Agent Claude Code

Adversarially reviews the change for security, privacy, authorization, abuse, dependency, and operational risks.

avivancos/open-english

Agent Claude Code

Challenges unnecessary complexity, duplication, premature optimization, and abstractions that make the implementation harder to understand or change.

avivancos/open-english

Agent Claude Code

Audits whether tests enforce the selected task's contracts, prohibit mocks, cover adversarial behavior, and fail when the implementation is wrong.

avivancos/open-english

Instructions file CodexOpenCode

AGENTS.md instructions for avivancos/open-english, covering agents.md — olcp (open language coach protocol), rule #0 — english only, project identity, commands and constraints before code.

avivancos/open-english

Instructions file

Claude Code instructions for avivancos/open-english, covering claude.md — adapter, claude code specifics and deploy triggers.

avivancos/open-english

Agent

This is the only file in this directory where a vendor model name appears.

avivancos/open-english

Agent

Everything in this directory and in .cursor/agents/ and .claude/agents/ is a copy. The source of truth is.

avivancos/open-english

MCP server Claude CodeCodexCursor

MCP server "olcp" as configured in avivancos/open-english. Runs locally from the olcp Python package.

avivancos/mcpforwork

Agent Claude Code

Internal documentation writer — runs after every card closes. Writes granular, module-scoped docs under docs/ capturing how the feature was developed (design decisions, invariants, files, tests, gotchas) so future agents become expert on the system. Also keeps existing docs accurate when a card changes documented…

avivancos/mcpforwork

Agent Claude Code

Fused review gate — audits that tests detect broken behavior (zero-mocks P0), reviews implementation correctness and contract fidelity, and runs a mandatory regression audit (full suite + structural gates + previous processes) so new work never breaks existing engineering.

avivancos/mcpforwork

Instructions file CodexOpenCode

AGENTS.md instructions for avivancos/mcpforwork, covering agents.md — source of truth for all agents and contributors, 1. product invariants (never violate), 2. architecture (pragmatic hexagonal), 4. testing contract (tdd, zero mocks) and 5. backlog workflow.

avivancos/mcpforwork

Instructions file

Claude Code instructions for avivancos/mcpforwork, covering claude.md — claude code complement, quick contract (the non-negotiables, condensed), claude code specifics and commands.

avivancos/mcpforwork

MCP server Claude CodeCodexCursor +2

MCP server "mcpforwork" as configured in avivancos/mcpforwork. Runs locally from the mcpforwork Python package.