cajias

Detect AI writing patterns in text. Use when: (1) need to check if text sounds AI-generated, (2) reviewing content before publication, (3) analyzing writing for Wikipedia-style AI markers, (4) generating report-only analysis without changes.

Use when: (1) User asks to "review for AI writing" or "check for LLM signs" (2) User wants to "detect AI-generated text" or "humanize this document" (3) User wants to identify and fix signs of AI/LLM-generated writing (4) User mentions Wikipedia AI writing guidelines

Fix AI writing patterns in text. Use when: (1) ai-writing-analyzer found issues to fix, (2) need to humanize AI-generated content, (3) preparing content for publication, (4) iteratively refining text until clean.

Humanize a document by removing AI-writing patterns via a multi-agent workflow (analyze → revise → adversarial review → loop-until-clean). Use when the user asks to "humanize", "de-AI", "remove AI tells", or "review for AI writing" on a file.

Fix for Claude Code slash command frontmatter that fails to parse when `argument-hint:` is given square-bracketed placeholder values. Use when: (1) authoring `commands/<name>.md` files in a CC plugin and the frontmatter shows `argument-hint: [working-dir] [team-name]` or similar multi-token bracketed values; (2) a frontmatter-validation test (e.g. `yaml.safe_load(block)` over the SKILL/command frontmatter) raises `ScannerError` or `ConstructorError` complaining about flow sequences or "expected a mapping"; (3) the slash command silently doesn't appear in `/help` after install, or a plugin-marketplace test asserting the `description` field is present trips because the whole frontmatter parse blew up. The fix is to QUOTE the value as a single string. Covers the same trap for `commands/<name>.md` directly authored in CC format AND APM-managed marketplaces shipping commands via APM.

Organize GitHub repository issues with proper milestones, native relationships, and clean labeling structure. Use when grooming issues, setting up milestones, establishing dependencies, or cleaning up issue tracking in a GitHub repository.

Enable Claude Code to delegate specific tasks to Amazon Q CLI agents running in the background with full tool permissions. Use when bridging to Q's specialized integrations such as Quip, ticketing systems, or diagram generation that Claude doesn't have direct access to.

Transfer markdown documents to Quip with proper formatting validation, handling common issues with lists, tables, and diagrams that occur during bulk uploads. Use when writing or syncing markdown content to Quip documents with section-by-section transfer for reliability.

Activate when a user invokes /agentic-harness:precommit, :review, :submit, or asks to run a Prose-packaged review workflow ("precommit review", "run mr-review on MR 21", "submit MR 42 via prose", etc.). Verifies prereqs, resolves reviewer defaults, and delegates to the openprose skill which embodies the Prose VM for the actual program execution. Not for one-shot questions or static documentation lookups — decline unless the user clearly wants to run a pipeline.

Rules for deduplicating code review findings across agents

Rules for auto-dismissing code review findings that are false positives

Local GitLab MR code review orchestrator. Use when the operator asks to "review this MR", "run a local MR review", "review my branch's MR", or wants a LOCAL-only (no GitLab posts) multi-dimensional review with automatic filtering and optional fix loop. Dispatches specialized review agents in parallel (intent, security, code quality, silent failures, type design, tests, comments), applies a 4-phase filtering pipeline, and iteratively fixes Critical/Important issues until convergence.

This skill should be used when the user asks to "address MR feedback", "fix review comments", "respond to code review", "handle MR suggestions", "shepherd an MR", "resolve discussion threads", or "run the MR feedback loop". Provides a systematic workflow for reviewing, implementing, and closing GitLab MR feedback — used standalone by operators AND as the thread-handling primitive for /agentic-harness:mr-submit.

Inspect and manage shepherd agents spawned by /agentic-harness:mr-submit. Use when the operator asks to "check MR status", "list active shepherds", "resume a shepherd after restart", "retire an MR", "approve drafted thread replies", or wants a dashboard of CI state and review-thread progress per MR. Reads state files under ~/.claude/mr-shepherds/; writes nothing except in response to an explicit mode flag.

End-to-end GitLab MR lifecycle orchestration. Use when the operator says "submit this MR", "open an MR with full review", "run the MR pipeline", "submit and shepherd", or wants to take a branch from "ready to push" through merge with automatic review, push, and CI/thread monitoring. Provides simplify → iterative review loop with TaskCreate-backed findings → push → open MR → background shepherd for CI + threads.

Activate when the user says "watch MR X until it lands", "babysit MR Y", "keep MR Z healthy", "keep an eye on this MR until it merges", or asks to run a long-running shepherd that survives across sessions. Complements `mr-shepherd` (single-pass) and `mr-submit` (end-to-end one-shot): mr-watcher is the persistent tail state that polls until GitLab marks the MR `merged` or `closed`, dispatching per-cycle shepherd work as new threads/CI events land.

Diagnose and fix failures in `paper.mode: live` runs against Binance Spot Testnet. Use when: (1) a live paper run aborts with `HttpClientBuildError("builder error")` from the Rust adapter; (2) a live paper run aborts with `MissingBinanceCredentialsError` from the fail-fast guard in `paper_phase._run_live`; (3) a live paper run reaches Binance and is rejected with `-2014` / `-2015` / `-1022`; (4) setting up a fresh Amazon Linux 2023 / RHEL / Fedora host for live paper for the first time; (5) the operator's shell exports a stale `SSL_CERT_FILE` (e.g., from isengard-cli) that points at a non-existent path; (6) a `compete status <run-dir>` post-mortem surfaces a live-paper failure root-cause bucket; (7) deciding which files in `nautilus-trading/` and the competition repo need to exist before running `compete run --paper-mode live`.

Operator runbook for the nautilus-competition CLI. Use when: (1) scaffolding a working directory via `compete init` (or laying it out manually); (2) running `compete run <working-dir>` with `--paper-duration-minutes`, `--rounds`, `--metrics-endpoint`, or `--run-id`; (3) reading the output tree under `<working-dir>/runs/<run_id>/` (eval/, paper/, leaderboard.md, final_leaderboard.md); (4) triaging a team that hit `max_train_iterations` and produced a `FAILED` marker / `FAILED.json`; (5) deciding between simulated and live paper modes; (6) running the hermetic demo at `examples/demo_competition/`.

Per-team authoring contract for nautilus-competition. A team is a MULTI-AGENT system (5-role baseline: researcher + hypothesis-generator + critic + risk-officer + memory-keeper, plus paradigm-specific specialists) that runs INSIDE `train(ctx)` to crystallize a strategy + `runtime_rules.json`. Use when: (1) creating or editing a team folder under `<working-dir>/teams/<team_name>/`; (2) designing the internal agent roster and orchestrating it in `train()`; (3) writing the researcher's `run_claude(...)` prompt or any paradigm-specific specialist stub; (4) writing or modifying a team's `entry.py` `def train(ctx) -> tuple[type[Strategy], StrategyConfig]`; (5) deciding how to use `ctx.prev_gain` for the anti-stuck paradigm-swap; (6) using `ctx.get_train_data` / `ctx.get_test_data` / `ctx.prev_round_leaderboard`; (7) designing the trade-time live Python critic that reads `runtime_rules.json` (no LLM at trade-time); (8) sizing the bar type vs `paper.duration_minutes` so indicators warm up; (9) respecting window disci

General-purpose NautilusTrader (nautilus_trader) usage guidance, OUTSIDE the nautilus-competition framework. Use when: (1) explaining NautilusTrader's event-driven architecture, MessageBus, or actor model for someone new to the library; (2) deciding between `BacktestEngine` / `BacktestNode` (historical replay) and `TradingNode` (live/testnet/paper over real venues) and describing the shared-Strategy-code principle; (3) writing a standalone `Strategy` subclass — lifecycle hooks (`on_start`, `on_bar`, `on_order_filled`), `order_factory` usage, indicator registration order; (4) picking `OmsType` (NETTING vs HEDGING) or `AccountType` (CASH vs MARGIN) for a non-competition backtest. Do NOT use this skill when (a) the question is specifically about the `nautilus-competition` repo (team contract, `compete` CLI, observability, leaderboard) — those have dedicated `nautilus-competition-*` skills; or (b) the user is hitting the logger-singleton panic or the catalog/instrument precision mismatch — those have dedicated `n

Live observability stack for nautilus-competition runs. Use when: (1) bringing up the docker-compose Prometheus + pushgateway + Grafana stack with `docker compose -f docker-compose.observability.yml up -d`; (2) wiring `compete run` to push metrics via `--metrics-endpoint` / `COMPETE_METRICS_ENDPOINT`; (3) the dashboard at http://localhost:3000/d/competition shows "No data"; (4) inspecting raw metric values via `curl http://localhost:9091/metrics`; (5) understanding the eight panels (Active Round, Eval gain_factor, Paper equity, Current price, Eval backtests, Latest metrics, Team failures, Avg eval duration); (6) wiping pushgateway state between runs.

Fix for "RuntimeError: invalid bar.volume.precision=N did not match instrument.size_precision=M" or similar bar/instrument precision-mismatch errors at BacktestEngine.run() time. Use when: (1) bars come from a ParquetDataCatalog and the instrument is loaded from somewhere else (TestInstrumentProvider, hand-built, hard-coded), (2) eval works but paper fails (or vice versa) on the same data, (3) you've seeded a catalog with one tool and are running the engine with an instrument from another. The fix is to load instruments FROM the catalog so their precisions match the bars stored alongside them.

Fix for nautilus_trader's global Rust logger panicking on second init within the same Python process. Use when: (1) you see "thread '<unnamed>' panicked at crates/common/src/ffi/logging.rs:146:14: Failed to initialize logging: attempted to set a logger after the logging system was already initialized", (2) a second `BacktestEngine` instantiation crashes mid-process even when the first one succeeded, (3) `Fatal Python error: Aborted` from a nautilus test after a prior test in the same pytest run already created an engine, (4) running `compete run` / any harness that creates multiple BacktestEngine instances per run aborts the second engine. Covers the bypass_logging fix, the pytest-process-isolation pattern, and why TradingNode → BacktestEngine sequences hit it too.

Rewrite, draft, and review prose so it is clear, specific, human, and appropriate to context. Use this skill when improving emails, essays, documents, reports, UI copy, marketing copy, posts, or any text that may sound generic, AI-written, over-polished, verbose, salesy, evasive, or structurally formulaic. Supports voice calibration, anti-slop audits, context-aware taste decisions, and final pre-flight checks.

Converts scripts, prose, story ideas, or existing AI image/video assets into professional cinematic direction deliverables: script breakdowns, director's book notes, continuity bible, shot lists, storyboard/keyframe prompts, image-to-video motion prompts, tool-specific prompt adapters, and QC repair notes. Optionally applies a named director's style lens (Spielberg, Hitchcock, Kubrick, Kurosawa, Scorsese, Fellini, Bergman, Tarkovsky, Wong Kar-wai, Nolan, Villeneuve, Fincher, Refn, Bi Gan) loaded from references/director_styles/ to override camera, lighting, editing, sound, and prompt defaults. Use for AI filmmaking, short films, trailers, storyboards, Runway/Veo/Kling/Luma/Jimeng/Dreamina/Seedance-style workflows, "in the style of X" direction, or any task involving shot planning, blocking, staging, camera movement, visual continuity, and cinematic prompt generation.

Use when planning a multi-shot AI video (TikTok Reel, Instagram Ad, YouTube Shorts, product explainer) where the target duration exceeds what a single AI generation can produce (>15s), and you need a coordinated shot list with visually consistent prompts for each segment

Generates optimized prompts for AI tools. Activates only when the user explicitly asks to write, fix, improve, or adapt a prompt for a specific AI tool (LLM, Cursor, Midjourney, image AI, video AI, coding agents, etc.). Does not activate for general conversation, coding tasks, document writing, or other non-prompt-engineering work.

Authoring contract for NautilusTrader strategies pluggable into the cajias/nautilus-trading `nt` CLI. Use when the user asks to "write a new strategy", "author a strategy for nautilus", says "STRATEGY_SPEC", "register a strategy", "external strategy", "build an algo for nt", or wants to port a backtest into the registry. Covers the StrategySpec / ImportableStrategyConfig / msgspec-frozen StrategyConfig contract, entry-point registration via pyproject.toml, plus ~10 canonical production gotchas spanning indicator scale, order time-in-force, multi-strategy actor pattern, monetary-string parsing, signal-handler safety, and CLI smoke-verification — full enumeration in the skill body.

Quickstart for the `nt` CLI shipped by the cajias/nautilus-trading repo. Use when the user asks "how do I run nautilus", mentions the "nt CLI", says "list strategies", "run a backtest", "paper trade with nautilus", "what does nt strategies do", "nt backtest", "nt paper-trade", "nt live", or is bootstrapping a fresh clone of nautilus-trading and needs the canonical workflow. Covers `make install`, `nt strategies` discovery via the `nautilus_trading.strategies` entry-point group, running backtests, paper-trading on Binance Spot Testnet, and a "discovery is empty" troubleshooting path. The public spec types live in `nautilus_trading.specs` (StrategySpec, ActorSpec, ConfigBuilder).

Be a live soccer coach for one team in the agentic-soccer match. Use when the user says "be the home coach" / "be the away coach" / "coach the home team" or wants to drive a team in the running soccer simulation via the soccer-sim MCP server. Self-paces a read→decide→override loop until full time.