sanurb

Scaffold or repair `etc/http/` OpenCollection trees for Bruno. Use for new collections, brownfield standardization, or Bruno/Postman migration.

Control herdr (a terminal-native agent multiplexer) from inside it. Manage workspaces and tabs, split panes, spawn sibling agents, read pane output, and wait for state changes — all via CLI commands that talk to the running herdr instance over a local unix socket. Use when running inside herdr (HERDR_ENV=1). Do not use outside herdr.

Sets up an `## Agent skills` block in AGENTS.md/CLAUDE.md and `docs/agents/` so the engineering skills know this repo's issue tracker (GitHub, GitLab, fp, or local markdown), triage label vocabulary, and domain doc layout. Run before first use of `fp-plan`, `fp-implement`, `fp-review`, `to-issues`, `to-prd`, `triage`, `diagnose`, `tdd`, `improve-codebase-architecture`, or `zoom-out` — or if those skills appear to be missing context about the issue tracker, triage labels, or domain docs.

Grilling session that challenges your plan against the existing domain model, sharpens terminology, and updates documentation (CONTEXT.md, ADRs) inline as decisions crystallise. Use when user wants to stress-test a plan against their project's language and documented decisions.

Find deepening opportunities in a codebase, informed by the domain language in CONTEXT.md and the decisions in docs/adr/. Use when the user wants to improve architecture, find refactoring opportunities, consolidate tightly-coupled modules, or make a codebase more testable and AI-navigable.

Build a throwaway prototype to flush out a design before committing to it. Routes between two branches — a runnable terminal app for state/business-logic questions, or several radically different UI variations toggleable from one route. Use when the user wants to prototype, sanity-check a data model or state machine, mock up a UI, explore design options, or says "prototype this", "let me play with it", "try a few designs".

Stress-test a plan against the domain model. Interrogates ambiguous terms, verifies claims against code, and updates CONTEXT.md, CONTEXT-MAP.md, and docs/adr/ inline. Do NOT use for greenfield ideation, code review, or AGENTS.md.

Deep codebase exploration across remote repositories. Fetch, search, and compare library source code. Use for internals, architecture analysis, and cross-repo patterns. Skip simple API or docs questions.

Create effective skills for OpenCode agents. Load FIRST before writing any SKILL.md. Provides required format, naming conventions, progressive disclosure patterns, and validation. Use when building, reviewing, or debugging skills.

Comprehensive Cloudflare platform skill covering Workers, Pages, storage (KV, D1, R2), AI (Workers AI, Vectorize, Agents SDK), networking (Tunnel, Spectrum), security (WAF, DDoS), and infrastructure-as-code (Terraform, Pulumi). Use for any Cloudflare development task.

Generate hierarchical AGENTS.md knowledge base for a codebase. Creates root + complexity-scored subdirectory documentation.

Create, update, transition, and manage Jira tickets. Use when working with Jira issues, tracking work, or automating ticket workflows.

Multi-repository codebase exploration. Research library internals, find code patterns, understand architecture, compare implementations across GitHub/npm/PyPI/crates. Use when needing deep understanding of how libraries work, finding implementations across open source, or exploring remote repository structure.

Dialogue-driven spec development through skeptical questioning and iterative refinement. Triggers: "spec this out", feature planning, architecture decisions, "is this worth it?" questions, RFC/design doc creation, work scoping. Invoke Librarian for unfamiliar tech/frameworks/APIs.

Detect whether the current project uses jj (Jujutsu) or git for version control. Run this BEFORE any VCS command to use the correct tool.

Implement work on issues. Use when user asks to "start working on issue", "what should I work on", "pick up task", "continue work", or "find next task".

Create and break down plans into trackable issues. Use when user asks to "create a plan", "break down feature", "design implementation", "structure this work", or pastes a Linear/GitHub/Notion URL to import.

Review code, create stories, and ensure commits are assigned to issues. Use when user asks to "review code", "assign commits", "check commits are assigned", "prepare for review", "create a story", or "write a story".

Review AI-generated code changes before committing using Ringi. Use to create review sessions, add comments/suggestions/todos, resolve decisions, export audit records, and (optionally) integrate via MCP.

Create effective skills for OpenCode agents. Load FIRST before writing any SKILL.md. Provides required format, naming conventions, progressive disclosure patterns, and validation. Use when building, reviewing, or debugging skills.

Proactively suggest diagrams when explaining complex systems. Triggers on diagrams, charts, visualizations, flowcharts, sequence diagrams, architecture diagrams, ER diagrams, state machines, Gantt charts, mindmaps, C4, class diagrams, git graphs. Use when user asks for visual representations of code, systems, processes, data structures, database schemas, workflows, or API flows. Generate Mermaid diagrams in markdown.

Convert markdown planning documents to Overseer tasks via MCP codemode. Use when converting plans, specs, or design docs to trackable task hierarchies.

Manage tasks via Overseer codemode MCP. Use when tracking multi-session work, breaking down implementation, or persisting context for handoffs.

Audit Effect v4 code for anti-patterns, type safety issues, and best practice violations using a deterministic multi-phase harness. Dispatches ripgrep for text-safe rules, ast-grep for structural rules, and reserves LLM judgment for semantic-only rules. Reports findings with file:line evidence, fix suggestions, and severity scoring. Use when reviewing, auditing, or checking Effect code quality.

Idiomatic Effect v4 TypeScript patterns. Auto-activates for Effect code: ServiceMap.Service, Schema.Class, TaggedErrorClass, Effect.gen/Effect.fn, Layer composition, branded types, @effect/vitest. Triggers on: Effect, @effect/, Schema, ServiceMap, Layer, TaggedError.

Generate Effect v4 boilerplate: services, schemas, errors, tests. Adapts to project conventions. Use when asked to create, scaffold, or generate new Effect structures.

Turn a plan, diagram, review, table, or explanation into a polished self-contained HTML page (Kami parchment design) to share in the browser. Use when asked to render, visualize, or "make this presentable/shareable", or proactively for any table with 4+ rows or 3+ columns. Outputs diagrams, slide decks, implementation plans, diff/code reviews, before/after comparisons, project recaps, data tables, and research explainers. Not for plain-text or markdown answers, terminal TUIs, backend-only code, or inline chat replies.

Designs and critiques data visualizations using Tufte's principles (data-ink ratio, chartjunk elimination, graphical integrity, small multiples, micro/macro, layering and separation). Use when the user asks to design a new chart, dashboard, or report; critique an existing visualization; reduce chartjunk; choose between viz approaches; plan sparkline placement, range-frame plots, slope charts, or layered annotations; or audit a graphic for lie factor. Produces a visual artifact via the zukai skill (HTML page with inline SVG or zukai primitives) when a before/after or redesign is warranted; replies in prose for purely conceptual questions. Don't use for general UI design, illustrations without quantitative content, brand-style decoration, marketing-page layout, or any task where the user is asking about typography/color/layout rather than data display.

Generate a self-contained handoff prompt for continuing work in a fresh session. Use when: "handoff", "hand off", "continuation prompt", "wrap up", "context transfer", "pick this up later", "pass to another agent".

Review code changes using Ringi — local-first structured review with web UI. Use when: "review", "code review", "/review", reviewing staged changes, branch diffs, or commit ranges. Ringi is the review tool; Pi is the workflow glue.

Control the cmux integrated browser for navigating, inspecting, and interacting with web pages. Use when asked to open, test, verify, fill, click, or read a web page.

Safe cmux workspace helpers for Pi — discover layout, create workspaces, select workspaces, open splits, focus surfaces, identify current context. All discovery returns JSON. Manages sidebar status and notifications automatically. Use when asked to: list or show workspaces, create a workspace, switch workspaces, split the view, check workspace layout, identify which workspace Pi is in, focus a surface, or interact with cmux layout. Triggers on: 'cmux', 'workspace', 'workspaces', 'split pane', 'new surface', 'open split', 'layout', 'which workspace', 'focus surface'. Does NOT support: sending text or keystrokes to terminals, reading terminal output, closing workspaces or surfaces, browser automation, or arbitrary cmux CLI passthrough.

Manage Asana tasks, projects, comments (stories), and activity via the asana-cli agent CLI. Use when: 'add a task', 'what's due today', 'check my tasks', 'complete a task', 'asana', 'task list', 'what did I finish', 'add a comment', 'daily review', 'search tasks', 'move task to project', 'what's overdue', or any task management request. All output is HATEOAS JSON with next_actions — parse result.tasks, result.comments, etc.