fakoli

This skill should be used AFTER code review and verification pass, when the user wants to "ship this", "ship the task", "open and merge the PR", "push, PR, wait for CI, and merge", or otherwise run the mechanical tail of a one-PR-per-task loop. It pushes the current branch, opens a PR, waits for CI to reach a terminal state, merges (squash + delete branch), syncs the base branch, and optionally runs a post-merge command (e.g. `anvil apply`). It makes NO review decisions — the human or agent gates that first.

Mine past coding-agent sessions (Claude Code, Codex, OpenClaw, Cursor CLI) into local-model eval suites executable via anvil-serving. Retro-first — consumes session-retro output dirs and cross-session findings themes, curates candidates into deterministic check-based evals sized for local models, emits anvil-serving-compatible eval-data suites, and runs them against any OpenAI-compatible endpoint. Use when the user asks to "create evals from my sessions", "build local-model evals", "turn this retro into evals", "which work classes can my local model handle", or wants evidence for anvil-serving routing decisions. Reads only local session logs; writes only to the eval-data root the user chooses.

Generate spec-INDEPENDENT breakage probes for a change — attack it along fail-closed, malformed-input, resource-exhaustion, and state-drift axes, not by re-reading its own tests. Use before opening a PR, as anvil execute's verify-left stage, when the user asks to "recall-mode verify", "red-team this change", "what breaks this", "find failure modes", or wants an independent breakage pass that doesn't reuse the implementer's assumptions. Reports findings; does not fix.

Scan source for Windows/cross-platform CLI hazards before shipping — non-ASCII in printed strings (the cp1252 console crash), hardcoded python3, heredoc backslash mangling, Node .cmd/.bat spawns, set -e in hooks. Use when the user asks to "check for encoding issues", "scan for Windows portability", "cli hygiene", is about to ship a CLI/script change, or wants the deterministic form of the ship-loop Windows discipline. Advisory (never blocks); wire it as a gate-router gate.

Route changed file paths to the verify commands this repo requires before shipping — docs changed means docs strict build, shell changed means bash -n, CLI changed means the encoding smoke test. Use before committing/opening a PR, when the user asks "what checks do I need to run", "run the gates", "gate check", or after substantive edits in a repo with a .claude/gate-router.local.md rules file. Deterministic local gates instead of session memory; also helps AUTHOR the rules file for a new repo.

Save or update this project's cross-session handoff note — the resume point for the next session, shared across checkouts of the same git remote and across linked worktrees. Use when the user types /handoff (optionally with a one-line summary), says "save a handoff", "note where we are for next time", "write a handoff before I clear context", or is wrapping up a session.

Show this project's saved cross-session handoff note (the resume point shared across checkouts of the same git remote and across linked worktrees). Use when the user types /recall, asks "where did we leave off?", "what's the handoff?", "what was I working on?", or "catch me up".

Run the full ship loop for a feature or fix — sync and isolate in a worktree, scope with ground-truth sweeps, implement with reality-faithful tests, gate on a multi-angle adversarial review, merge, and close the follow-up loop (promotions ledger, out-of-diff issues). Use when the user says "ship this the usual way", "run the loop", "ship-loop", starts substantive feature/fix work in a fakoli repo, or asks to take a change from idea to merged PR with the standard review gate. The adversarial review is the merge gate — never merge substantive work without it.

Author, parse, and review a project PRD in anvil — capture the requirements that everything downstream (features, tasks, claims, evidence) gets generated from. Use this skill when starting a new project or revising requirements before any planning work happens.

Decide what to do with an anvil task that has submitted evidence and is awaiting human review — accept and ship, reject and reopen, or hold for further investigation. Use this skill when one or more tasks are in needs_review and need a final disposition.

Turn a reviewed PRD into a ready-to-execute task graph — generate features and tasks, score each on six dimensions, surface dependencies and conflict groups, promote drafted tasks to ready. Use this skill once the PRD is approved and before any agent claims work.

Bootstrap a `.anvil/prd.md` draft from a rough project idea — interview the user question-by-question and write the result so `anvil prd parse` can consume it. Use this skill when the user has a project intent but does not yet have a PRD (e.g., asks to "start a PRD", "draft requirements", "author a PRD", or "spec out a project").

Run the agentic execution loop on a claimed anvil task — fetch the work packet, do the work, submit completion evidence. Use this skill when an agent has just claimed a task and needs to execute it end-to-end without juggling individual CLI commands.

Inspect anvil project state — list tasks, show task details, find the next claimable task, summarize active claims and blockers, check file-conflict warnings, and reconcile state with the filesystem and git. Use this skill when you want to see what anvil knows without changing anything.

Acquire an exclusive lease on an anvil task — pick from the ready queue, check for file conflicts, claim the task, and get a working git branch to commit into. Use this skill when ready to start work on an approved task.

Walk the PRD's unresolved items — `[NEEDS DECISION]` markers, `## Open Questions`, and missing acceptance-criteria or verification fields — and drive each one as a Q&A turn with the user, proposing concrete options when possible and applying the chosen answer with `anvil prd resolve-decision`. Use this skill when `anvil prd find-decisions` reports unresolved items, or when other skills (prd, plan) detect decisions blocking progress.

Choose how to handle a new coding request — native for tiny edits vs a fakoli-claw flow for multi-file, risky, design, or explicit crew/Fakoli work.

Design phase — refine an idea into an approved spec through structured one-question-at-a-time dialogue, propose 2-3 approaches, present the design section by section, write the spec, and hand off to flow:plan. Works headless over any channel.

Execute phase — wave-based crew dispatch on OpenClaw with critic gates and evidence-based verification. Load an intent-driven plan, group tasks into dependency waves, sessions_spawn specialists in parallel per wave (local SGLang tier), run a critic gate after every code wave, then a sentinel evidence gate.

Ship phase — re-run fresh tests, then present exactly four options (merge locally, push+PR, keep as-is, discard) and execute only the explicit choice. Never auto-merge, auto-push, or discard without the typed word discard.

Plan phase — break an approved spec into an intent-driven task list for OpenClaw crew execution. Verify assumptions with a scout sub-agent, map files to owners, write WHAT-not-HOW tasks with acceptance criteria and Depends on, then hand off to flow:execute.

Fast path — skip brainstorm/plan/waves for small tasks under 3 files. Estimate scope, dispatch a single specialist (welder by default), run verification, run a one-cycle critic gate, done. Escalates to flow:execute if scope or fixes exceed the limit.

Verify phase — evidence-based validation. Detect language, run the full check suite, dispatch the sentinel against the plan's acceptance criteria, and produce a pass/fail scorecard where every PASS cites fresh command output from this session.

Manage the Fakoli Style operating-model principles ledger — add principles, advance lifecycle status, validate the ledger, and read the current status report. The canonical source is data/principles.json; docs/fakoli-style.md is a generated projection.

Earnings season preparation covering upcoming earnings dates, EPS estimates, historical beat rates, pre-earnings technical setups, and portfolio risk assessment. Use this skill to prepare for earnings announcements across your portfolio holdings. Ask for an "earnings preview", "upcoming earnings", "prep for earnings", "earnings season overview", "earnings calendar", "who reports this week", or "earnings this week" to trigger this workflow. Produces an Earnings Preview contract output.

Comprehensive portfolio health check covering performance, risk metrics, asset allocation, RSU exposure, and benchmark comparison. Use this skill when you want to review your portfolio, check diversification, assess risk, or understand RSU concentration. Ask "how is my portfolio", "portfolio health", "portfolio review", "check my portfolio", "portfolio performance", "my holdings", or "am I diversified" to trigger this workflow. Produces a Portfolio Scorecard contract output.

Investment decision support for a specific stock — builds an investment thesis with valuation, technical entry points, portfolio fit analysis, position sizing, and risk management. Use this skill when deciding whether to buy, sell, or hold a position. Ask "should I buy AAPL", "should I sell NVDA", "trade idea for AMZN", "evaluate PINS", "investment thesis for MSFT", "is it a good time to buy GOOG", "trade research on META", or "add to my position in TSLA" to trigger this workflow. Produces a Trade Decision Brief contract output.

Deep-dive research report on a single company covering financials, technicals, analyst ratings, news catalysts, and portfolio fit. Use this skill when you need to research a stock, perform due diligence, or want a comprehensive company analysis. Ask to "research AAPL", "deep dive into NVDA", "analyze company PINS", "stock analysis", "tell me about MSFT", "due diligence on AMZN", or "look into GOOG" to trigger this workflow. Produces a Research Report contract output.

Comprehensive macro economic analysis covering business cycle positioning, inflation trends, labor market conditions, yield curve signals, Fed policy outlook, and portfolio implications. Use this skill when you need to understand the economic environment, assess recession risk, or evaluate how macro conditions affect your portfolio. Ask for an "economic outlook", "macro analysis", "fed watch", "recession risk assessment", "economy overview", "macro outlook", "interest rates analysis", "yield curve check", or "inflation update" to trigger this workflow. Produces an Economic Brief contract output.

Generate a comprehensive morning market briefing covering index performance, sector rotation, macro indicators, and portfolio impact. Use this skill when you need a broad market summary, want to know how markets are doing, or need a morning briefing. Ask for a "market overview", "morning briefing", "market summary", "daily briefing", "what's the market doing", or "what happened in markets" to trigger this workflow. Produces a Market Briefing contract output.

Targeted review of failure modes, hidden dependencies, limits, and operational burden in a specific architecture or design. Use this when you need a focused risk assessment without running a full complexity mapping — to identify single points of failure, blast radius concerns, hidden coupling, and operational survivability gaps. Produces a Hidden Risk Summary with dependency risk matrix, compound failure scenarios, and prioritized remediation actions. Ask for an "architecture risk review", "failure mode analysis", or "dependency risk assessment" to use this workflow.

Surface hidden complexity, risks, dependencies, and likely project blow-up points in a proposed design or architecture. Use this when you have architecture docs, design proposals, or technical specs and need to identify where operational, cost, networking, or implementation complexity is hiding before committing resources. Produces a Complexity Heat Map and Hidden Risk Summary with source-anchored evidence and severity ratings. Ask for a "complexity map", "hidden risk analysis", "blow-up point review", or "complexity audit" to trigger this workflow.

Split large or varied source material across parallel subagents for extraction before synthesis. Use this when documentation exceeds what can be processed in a single pass — typically more than 5 dense documents or 50+ pages — or when material spans multiple distinct domains. Produces structured Context Packets that can feed into complexity-mapper, decision-brief, or architecture-risk-review. Ask to "shard documents", "split for parallel extraction", or "break this into digestible chunks" to use this workflow.

Package extracted findings into a compact, decision-ready format for engineers, managers, or review bodies. Use this after running complexity-mapper, architecture-risk-review, or context-sharding to synthesize findings into a structured brief with options, evidence, risks, and recommended next steps. Produces a Decision Brief with evidence summary, top risks, unresolved questions, and actionable recommendations. Ask for a "decision brief", "decision package", or "summarize findings for a decision" to use this workflow.

Generate high-quality first drafts by adapting prior proven work to new problems. Use this when you have successful prior designs, implementations, or architectural patterns and need a strong starting point for a new but related problem. Produces a Pattern Remix Draft that maps reused elements to their sources, flags carried assumptions, identifies divergence risks, and provides implementation steps. Ask to "remix a prior design", "adapt existing work", or "create a first draft from prior patterns" to use this workflow.

Operate anvil-serving from an agent harness using MCP/controller-backed playbooks, CLI fallbacks, and sub-agent evidence packets. Use when asked to inspect Anvil Serving status, choose models, run preflight or benchmarks, sync OpenClaw/Codex/Claude Code workbench config, collect promotion evidence, or coordinate small-model sub-agents for anvil-serving operations.

Operate anvil-serving from an agent harness using MCP/controller-backed playbooks, CLI fallbacks, and sub-agent evidence packets. Use when asked to inspect Anvil Serving status, choose models, run preflight or benchmarks, sync OpenClaw/Codex/Claude Code workbench config, collect promotion evidence, or coordinate small-model sub-agents for anvil-serving operations.

Operate anvil-serving from OpenClaw using MCP/controller-backed playbooks, CLI fallbacks, and sub-agent evidence packets. Use for router status, model selection, preflight, benchmark probes, OpenClaw sync, promotion evidence, and safe small-model workflow slices.

Operate topology-owned Anvil Voice lifecycle, Mini-to-Dark audio routing, sidecars, profiles, proxies, and bounded voice benchmarks. Use for voice status, logs, start/stop/restart, OpenClaw Talk checks, and voice evidence.

Operate topology-owned anvil-serving voice audio and Realtime proxy lifecycle, Mini-to-Dark loopback forwarding, profiles, sidecars, and bounded voice benchmarks. Use for voice status, logs, start/stop/restart, OpenClaw Talk topology checks, and voice evidence that must remain separate from router promotion evidence.

Inspect recently changed or explicitly targeted code for smells, cleanup opportunities, language idioms, simplification, and safe source shrinking while preserving behavior. Use when the user asks to clean up, simplify, shrink, refine, make clearer, polish, reduce complexity, check code smells, or review recent code changes for maintainability. Build a bounded hygiene packet, consult the bundled smell/examples references, delegate to the focused hygiene-worker when available, edit only scoped files, and verify the result.

Focused worker for Code Hygiene packets. Use only when a hygiene packet or explicit scoped file list is provided. Apply small behavior-preserving cleanup, simplification, idiom, or safe shrink edits using the bundled smell and example references, stay within scope, and return verification evidence.

Read cached Code Hygiene observations and produce a read-only findings report. Use when the user asks for cached smells, a hygiene report, or a non-editing summary of current findings.

Guiding principles for AI-assisted Python backend development. Use when building APIs with FastAPI, designing distributed systems, implementing microservices, or working on any Python backend code. Provides architectural patterns, code standards, testing strategies, and AI workflow integration. Apply when (1) creating new backend services or APIs, (2) refactoring existing Python backend code, (3) designing system architecture, (4) implementing database patterns, (5) writing tests for backend code, or (6) reviewing AI-generated backend code.

Design reasoning and UX principles for frontend interfaces and design documentation. Use automatically for UI or frontend design work, and when explaining or critiquing design decisions, accessibility, or design system choices, or when the user mentions @frontend-design-principles.

End-to-end testing workflow for nat464-sidecar in IPv6-only Kubernetes clusters. Use when setting up test environments, deploying the sidecar to k3s, verifying IPv6-to-IPv4 translation (inbound and outbound), benchmarking with iperf3, or troubleshooting pod networking issues. Triggers: 'test the sidecar', 'set up test cluster', 'verify IPv6 translation', 'deploy to k3s', 'benchmark sidecar', 'test inbound/outbound', 'IPv6-only cluster setup'.