claude

experimental

Skills still proving themselves — behavior and interface may change.

1 on GitHubMITupdated 6 days ago
/

Plugin experimental from marketplace furanku-skills in RafaelVidaurre/furanku-skills. Commands read from the repo's manifest — 5 days ago.

What it installs

Skills (15)

  • agent-readable-logs

    Establish, audit, and maintain logging that coding agents can retrieve through documented bounded commands - one stable retrieval interface per emitter, one parseable event per line with correlation fields, actionable failure summaries from project-owned scripts, and a scoped Logs block in the agent

  • codemap

    Build and maintain an architecture map of a codebase as a standalone HTML explorer with five lenses (Purpose shows people, areas, runnables, and flows; Layers shows where each part runs and which way dependencies point; Ships in shows which libraries each app, service, and CLI contains; Size & activ

  • codexbar

    Read Claude and Codex account quotas through the official CodexBar CLI. Use when checking usage or rate-limit failures, setting up multiple account profiles, installing CodexBar CLI, or migrating an older quota reader to CodexBar.

  • council

    Convene a council of AI models (Claude, Codex, Grok, Gemini, ...) that independently take positions, challenge each other, and vote their way to consensus. Use when the user says "council", wants several models to debate or weigh in on a decision, asks for multiple model perspectives, or wants the b

  • crew

    Establish Commander, Captain, and Worker ownership over delegated work on any orchestration mechanism and any issue tracker. Use when the user asks an agent to act as or become Commander, Captain, or Worker; asks to spawn or delegate a Captain or Worker; asks Commander to take command of a Captain;

  • decision-trail

    Keep a reviewable, append-only TSV trail of consequential decisions, reasons, evidence, and results. Use when the user requests a decision trail, or when substantial multi-phase work will be reviewed after the run.

  • guidance-composer

    Compose selectable engineering guidance from a curated catalog and inject it into a project via the furanku-skills guidance-composer CLI namespace (interactive or non-interactive) or by answering catalog questions from references/catalog.json. Use when the user wants to inject guidance, compose proj

  • human-playtest

    Record audio and video while a person speaks feedback about an app, website, prototype or game, with timestamped screenshots. Use to capture a feedback session, control an ongoing recording, or extract and organize feedback from supplied evidence.

  • model-routing

    Choose an agent, model, and effort using capability evidence, routing preferences, and live quota, then gate-check the launch decision. Use before delegating work; to list accounts and usage quotas, configure account priorities, list models, or set a model and effort enabled, disabled, or explicit;

  • progress-report

    Create evidence-backed project progress reports with current status and non-overlapping time bands, recording the underlying dated facts for future runs. Use only from an unspawned root agent when the user requests a progress report, or automatically when a formal harness goal completion or meaningf

  • promo-music-video

    Produce a product promo music video from a song, built on reviewed lyric and beat timing, real product footage, kinetic typography, and music-driven motion. Use when asked to turn a song into a product video, check lyric sync before designing visuals, or polish an existing music-led promo.

  • prr

    Use and maintain a Product Requirements Record in repos containing prr/. Read it before planning, implementing, reviewing, or evaluating product behavior; capture new durable product intent from conversation. Use on request to start tracking requirements, resolve tentative requirements or open quest

  • testing-best-practices

    Guides writing, running, reviewing, and pruning automated tests so each test is risk-proportionate evidence of general behavior and each run covers only what changed. Use when writing or reviewing automated tests, deciding whether a change needs a test, choosing test scope or doubles, deciding which

  • usage-drain

    Report which model and effort levels are draining Codex, Claude, and Grok usage from local session logs. Use when the user asks what is draining usage, and when they ask for a usage breakdown since the last reset or over a stated time range.

  • winremote-host-runs

    Run project workloads and transfer artifacts through winremote MCP with durable ownership and conflict-aware concurrency. Use for Windows-host tests, builds, imports, captures and benchmarks, coordinating local runs, or setting up, connecting and upgrading the host-runs winremote server and client.

View source on GitHub

Maintain experimental?

This page updates itself from your repo. Add the badge to your README, or get it Featured to pin it above the organic results with a followed link to your site.

claude & codex pluginslisted