Claude Skills by isvlasov

Delegates a task to OpenAI Codex running as an interactive session in a herdr pane - uses the user's ChatGPT subscription, visible in herdr, steerable mid-session. Use when delegating work to codex, offloading a task to a GPT model (sol, terra, luna), or running a second opinion from a non-Claude frontier model. Not for Claude Code's own Agent tool or local-LLM delegation (delegating-to-local-llm).

Delegates a task to a local LLM running as a Pi coding-agent session in a herdr pane - the subagent is visible in herdr, can be steered mid-session, and costs no cloud tokens. Use when delegating work to a local model, spawning a Pi subagent, offloading a task to omlx (qwen, gemma), or running agent work offline. Not for Claude Code's own Agent tool or subagents inside Pi.

Loads the rageatc-code-oss operating rules directly from the plugin when the SessionStart hook has not delivered them. Use when the rageatc-code-oss failsafe line appears in session-start hook output, when the plugin's operating rules are absent from context, or when the injected rules look truncated.

Sweep the project's LEARNINGS.md inbox and turn accumulated entries into durable skill, rule, or documentation changes. Run when ready to process learnings — not mid-task.

Raises an artefact's signal-to-noise ratio by removing text that earns nothing for its reader — filler, repetition, over-explanation, ornamental phrasing — and relocating conditional detail to references. Use to slim a skill, agent, or document, cut bloat, tighten prose, or shrink a file that has grown too long. Tuned for SKILL.md files.

Designs software architecture from a confirmed PRD. Use when a PRD exists and architecture must be designed before implementation, when writing ADRs, choosing an architecture style, or selecting technology and databases.

Writes correct, version-aware Telegram bot code. Use when writing, extending, or debugging a Telegram bot in python-telegram-bot, aiogram, grammY, or Telegraf. Not for Telegram client API (TDLib), languages other than Python and Node.js, or non-Telegram platforms.

Converts an approved ARCHITECTURE.md into an implementation roadmap of isolated, dependency-ordered chunks. Use when architecture has been approved and work must be decomposed before implementation begins.

Enriches implementation roadmap chunks with acceptance criteria, pattern references, and contextual notes. Use after decomposing-work has produced the structural roadmap and before planning-software creates the orchestration plan. Does not modify chunk structure or dependencies.

Evaluates software quality from an end user's perspective. Use when emulating a real user to assess a running web app, CLI tool, or API.

Retrieves up-to-date documentation, API references, and code examples for any developer technology. Use whenever the user asks about a specific library, framework, SDK, CLI tool, or cloud service — including well-known ones like React, Next.js, Prisma, Django, or Spring Boot. Use for API syntax, configuration, version migration, setup, and library-specific debugging. Use even when you think you know the answer — training data and web search are frequently outdated for API details.

Extracts product requirements and produces a PRD. Use when a user wants to build software and needs to define what to build before architecture or implementation, when scoping an MVP, or when translating a vague idea into a buildable specification.

Runs software development projects from idea to completion. Use when starting a software project, adding a feature, fixing a bug, or resuming an in-progress build.

Transforms an approved enriched roadmap into a self-contained orchestration plan. Use when creating an orchestration plan, producing a build plan, or resuming a project from a saved plan. Not for roadmap creation/enrichment or workflow tier selection.

Reviews code. Use when reviewing code implementations, assessing code quality, running a security review, checking spec or design system compliance, or conducting an end-of-build whole-project review against PRD success criteria.

Creates the standard project file structure for a software project. Use when starting a new software project, initialising project files, or setting up a project workspace before architecture and decomposition begin.

Guides root cause investigation before proposing fixes. Use when debugging, diagnosing failures, investigating test errors, or tracing unexpected behaviour.

Enforces the red-green-refactor TDD cycle. No production code without a failing test first. Use when implementing features, fixing bugs, or writing production code.

Manages git worktree isolation for implementation chunks. Use when setting up isolated workspaces, creating worktrees for chunks, or merging completed chunk branches.

Gates completion claims with mandatory evidence — no "should work" or "probably passes" without fresh verification output. Use when verifying test results, confirming builds pass, checking linter output, or validating acceptance criteria.

Conducts quality reviews of non-code artefacts. Use when reviewing or critiquing Skills, agents, workflows, documentation, or written materials, assessing quality against standards, or providing structured feedback.

Extracts learnings from completed work. Use when capturing lessons learned, analysing what went wrong or what went right, conducting root cause analysis, or creating improvement proposals for skills or workflows.

Collects research sources with quality evaluation. Use when gathering or finding sources, building a source library, searching for academic papers, or performing RADAR assessment.

Guides written communication — emails, reports, presentations, documents, website copy, newsletters, books. Use when writing or improving writing, adapting to an audience, or shaping document structure. Not for oral delivery or tool-specific formatting (slides, PDFs).

Conducts research. Use when asked to research, investigate, explore, look into a topic, compare options, learn about an unfamiliar domain, or understand current state.

Transforms unstructured requests into clear, actionable briefs. Use when writing a brief, preparing a specification, extracting requirements, or clarifying what a task needs.

Invokes the critic-agent to review the latest produced output. Use when the user runs /critic to request a quality review of the most recent artefact or deliverable from the current workflow.

Designs workflow structure. Use when planning a workflow, choosing which phases to include, or selecting a workflow pattern for a task.

Re-explains concepts at a "knows the basics" level — strips jargon, uses analogies, and assumes broad general knowledge without domain expertise. Use when the user invokes /eli40 to request a simpler explanation of the previous message, or /eli40 <topic> to get an accessible explanation of a specific topic. Like ELI5 but calibrated for an adult generalist.

Run /handover then commit and push changes to git

Session handover — end a session by systematically updating project state files for continuity. Use this when the session is ending, the user asks for a "handover", or when significant progress needs to be recorded.

Generates and explores ideas. Use when capturing fleeting thoughts, brainstorming, exploring a problem space before committing to a direction, or reviewing stored ideas to decide what to pursue.

Capture an observation — something that went wrong or worked well — to LEARNINGS.md for future reference. Quick in-session capture, not full analysis.

Principles for multi-agent workflow execution. Use when coordinating agents, accepting or rejecting agent work, executing producer-critic-learner cycles, or running any multi-phase orchestrated task.

Extracts genuine value from a subject and frames it so the right audience recognises it. Use when producing LinkedIn posts, pitch narratives, cover letters, proposals, case studies, or positioning statements. Not for general writing guidance, negotiation, or copywriting templates.

Session prime — start a fresh session by reading project state to orient yourself. Use this when you have no project context yet or the user asks to "prime the session", "read state", or start work.

Challenges the user's reasoning, assumptions, or proposals — a thinking sparring partner that pokes holes without the full critic workflow. Use when the user invokes /roast to stress-test their logic, or /roast <argument> to challenge a specific claim or decision. Not a review of artefacts — a challenge to thinking.

Shape an idea into a concept through guided dialogue

Explores solution options when the problem is defined but the approach is unclear. Use when evaluating approaches, comparing alternatives, or choosing between options.

Agent behavioural frame for operating within rageatc. Preloaded by agents — not for orchestrator use; orchestrator context comes from rules.

Clarifies user intent before formalising requirements. Use when requests are vague, incomplete, or solution-focused, or when preparing to write a brief.

Verifies claims and assesses source credibility. Use when fact-checking claims, verifying accuracy, validating sources, detecting hallucinations, triangulating evidence, or performing SIFT or RADAR assessment.

Creates and updates Claude Code Agent Skills. Use when creating a skill, revising SKILL.md, improving a skill's name or description, or adding progressive disclosure files.

Creates a design system for software with a UI. Use when a project has a user interface and architecture is confirmed — whether creating from scratch or extracting patterns from existing code.

Automates browser interactions. Use when navigating websites, interacting with web pages, filling forms, taking screenshots, testing web applications, or extracting data from web pages.

Handles PDF operations (reading, creating, modifying). Use when extracting text from PDFs, converting markdown to PDF, merging or splitting PDFs, compressing, rotating pages, handling metadata, or when Claude Code's native PDF reading fails.