flowai-plugins

Write or fix a prompt for a fast/cheap model (Gemini Flash, GPT-4o Mini, Haiku) so its output is stable and parseable. Use for any request to draft, tune or debug such a prompt. Not for reasoning models (use engineer-prompts-for-reasoning).

Write or fix a prompt for a reasoning model (Gemini Pro, GPT-4o, Claude Sonnet) so it holds context across a multi-step task. Use for any request to draft, structure or debug such a prompt. Not for fast models (use engineer-prompts-for-instant).

Guide for creating persistent AI rules — coding standards, project conventions, file-specific patterns. Use when users want to create a rule, set up conventions, or ask where rules go. Works across IDEs (Cursor, Claude Code, OpenCode).

Guide for creating agent-invocable Agent Skills (SKILL.md packages auto-discovered by description). Use when users want to create or author a skill, or ask about skill structure or format. Not for user-invoked slash commands. Works across IDEs.

Guide for creating effective Subagents (custom agent definitions). Use when users want to create a new subagent, set up task-specific agents, configure code reviewers, debuggers, or domain-specific assistants. Works across IDEs (Cursor, OpenCode).

Use when the user asks to plan a LARGE feature spanning multiple sessions or phases — produces an epic under the `tasks` role with dependency-ordered phases, atomic tasks and per-phase status. Use plan for single-session tasks, not for roadmaps.

Use when the user asks to execute an existing task plan's Solution under TDD — RED → GREEN → REFACTOR → CHECK per AGENTS.md. Requires a written plan under the `tasks` role. Do NOT trigger on planning, review, or unrelated failing tests.

Use when the user asks to set a project up for flowai for the first time — writes AGENTS.md and the documentation skeleton for a new or an existing codebase. Not for reconciling an existing AGENTS.md with newer templates.

Creates interactive HTML teaching materials with clickable state diagrams. Use when the user asks to produce an explorable tutorial artifact, not to answer a question.

Use when the user asks to diagnose a bug, find the root cause of a failing behavior, or run a hypothesis-by-hypothesis investigation with experiments and evidence. Produces a diagnostic report with a recommended fix, but does not apply the fix.

Use when the user asks for a project health audit, a maintenance sweep, or a multi-category lead-engineer scan followed by interactive issue-by-issue resolution. Not for routine lint/test runs, single-file cleanups, or plain check-project requests.

Use ONLY when the user explicitly asks to plan a task, create a task file, produce a task breakdown in the accepted format, or prepare a critiqued strategy before coding. Writes under the `tasks` role. Not for design chat or ad-hoc code suggestions.

Use when the user asks to push the current branch — confirms a first-time upstream and a diverged main, and refuses force pushes without explicit authorization. Not for committing or reviewing. Self-contained — execute the inlined steps directly.

Use when the user asks to reflect on the CURRENT session or one transcript file they name, analyze the agent's own process, audit context use, or extract lessons from how a task was done. Not for multi-session history — use reflect-by-history.

Use when the user asks to review PAST sessions, find recurring patterns across sessions, or improve project primitives (rules, skills, hooks, docs) from historical IDE transcripts. Do NOT trigger on current-session self-review — use reflect instead.

Use when the user asks to close a session by auditing it AND writing the fixes into the project's instruction files — audits, applies and commits the edits, then asks whether to push. Not for a reflection that only reports findings.

Use when the user asks to review CURRENT uncommitted changes (staged, unstaged, branch diff) before committing, or to synthesize JiT tests against the diff — QA + lead-engineer verdict on quality, architecture, cleanup. Needs a diff, not old commits.

Use when the user asks to review the uncommitted changes and then commit them — the verdict gates the commit. Not for a review alone or a plain commit. Self-contained — execute the inlined steps directly, do NOT invoke other skills.

Use when the user asks to review finished uncommitted work, commit it and push it in one run. Not for planning or implementing, nor when the user stops before the push. Self-contained — execute the inlined steps directly, do NOT invoke other skills.

Use when the user gives a source (URL, file, or text) to save into the project's memex — a long-term knowledge bank for AI agents. Stores the raw source, extracts cross-linked entity pages, audits backlinks, updates the index. Not for casual reads.