最近新增
刚加入收藏的新技能，抢先试用。
plan
ZachArticulateV
"Shape and plan software work: settle requirements through an interview, make a judgment call with evidence, write an implementation-ready plan, or split a plan into dependency-ordered tickets. Use when the user wants to plan a feature, think through requirements, choose between approaches, or break work into tasks or tickets."
learn
ZachArticulateV
"Make the project and its agents learn from experience: capture a lesson from work just done, turn a repeated mistake into a check that prevents it, run a retrospective on a session, refresh stale lessons, or update the user's working profile from their history. Use when the user says to remember a lesson, says an agent keeps making the same mistake or corrects one, asks for a retro, or asks agents to work the way they do."
debug
ZachArticulateV
"Find the root cause of a bug and fix it with proof: reproduce the failure, test competing explanations, trace the cause from trigger to symptom, and fix it behind a regression test. Use when something is broken, failing, flaky, slower than it used to be, or behaving differently than expected."
catch-up
ZachArticulateV
"Catch the user up: what changed in this conversation since their last message, or where a piece of work stands across sessions, branches, and PRs, ending with the decisions that wait on them and the next move. Use when the user asks to be caught up, for a status or summary of what happened, where they left off, or what they were working on. For the health of the memory tree itself, use project-memory status."
build
ZachArticulateV
"Implement planned work with proof at every step: execute a plan or tickets unit by unit, running independent units in parallel; refine code just written; or improve a measured metric, keeping only the wins that clear the noise. Use when the user asks to implement, build, execute a plan or ticket, clean up freshly written code, or make something faster or smaller."
autopilot
ZachArticulateV
"Take a request from idea or bug report to a reviewed, verified pull request without stopping, by running this plugin's skills in order: plan or debug, build, simplify, verify, review, learn, and ship. Use when the user asks for hands-off, end-to-end, or autonomous work, such as \"take it all the way\" or \"just do it\"."
triaging-tester-reports
chocoliticekreem
Use when a playtester, teammate or AI tester report describes bugs in a Roblox game in loose words ("hose sometimes doesn't spray", "my buddy got blamed", "horn too quiet") and has to become fixes, or when a test_feature/playtest/swarm report lists several bugs.
training-mubit-from-corrections
chocoliticekreem
Use when an AI playtester, Forge role (fixer, auditor, critic, triage) or AI player made a wrong call and the user wants Mubit to learn from it ("make it learn", "correct the AI", "this should be a training example"), or when preparing the demo's correction → model update moment.
testing-roblox-features
chocoliticekreem
Use when asked to check, verify or prove that a Roblox gameplay change or feature works in Studio ("does the hydrant refill work now?", "check my change", "test X before the demo"), or before saying a game change is done.
swarm-testing
chocoliticekreem
Use when many testers should hit a Roblox game at once ("throw 10 testers at it", "stress the lobby", "parallel AI testers", load or concurrency bugs in squads, pads, countdowns or mission instances).
mubit-tester
chocoliticekreem
Use when checking that a Roblox game change actually works in play, reproducing a gameplay bug, testing multiplayer, co-op or squad behaviour (lobby, pads, countdown, parallel mission instances, friendly fire, revive, hose tether), or running playtests/scenarios against Roblox Studio with the mubit-tester MCP tools.
improving-games-until-done
chocoliticekreem
Use when an existing Forge game (runs/<slug>) should keep improving until it is ready ("keep improving until forge done is green", "the tycoon is scuffed, fix it", "make it good"), or when deciding whether a Forge game is finished.
forge
chocoliticekreem
Use when driving the forge MCP tools (build_game, improve_game, done, correct, tester_report, versions, rollback, jobs) on Forge-made Roblox games, or checking how Forge's self-improving loop (auditor, fixer, verifier, Mubit training) is doing.
building-games-with-forge
chocoliticekreem
Use when the user wants a new Roblox game made from an idea ("make me a co-op lava escape", "build a tycoon with Forge"), especially when they will be away while it builds.
roblox-studio-dev
chocoliticekreem
Use when writing, changing, debugging or playtesting Luau game code in this repo (game/ or runs/<slug>/game), when Roblox Studio or mcp__Roblox_Studio__* tools are involved, or when a change has to be checked inside a running Roblox game.
forge-orchestrate
chocoliticekreem
Use when the user wants Forge to make an existing game much better on its own (for example "make the tycoon 10x better"), to find what a game is missing or what is broken, to queue several Forge jobs that share Studio windows, or to correct the gap list Forge produced. Drives forge orchestrate, forge gaps and forge queue (or the forge MCP tools orchestrate, gaps and orchestrator).
forge-autopilot
chocoliticekreem
Use when the user wants Forge to take a game from an idea to ready on its own, to keep improving a game driven by its logs and playtest evidence, to find anomalies (error spikes, counters that never move, regressions, stuck testers, features nobody uses), to get a read-only improvement report on the human-built main game, or to correct an autopilot decision. Drives forge autopilot (or the forge MCP tool autopilot).
statistical-red-team
takenthatusernameis-collab
Audit leakage, survivorship, overlapping horizons, sparse crash tails, confounding, multiplicity and regime dependence. Use before accepting an effect or changing a hypothesis; never use a sealed holdout for parameter search.
resumable-research-operations
takenthatusernameis-collab
Execute long-running quantitative research with immutable IDs, bounded retries, complete cell ledgers and durable handoffs. Use for every session; never promise uninterrupted uptime or suppress failures.
point-in-time-universe
takenthatusernameis-collab
Build an auditable candidate frontier ordered by descending market-cap rank and map candidates to historical perpetual contracts and spot/index series. Use for asset discovery, rank history, listing/delisting, or mapping; never use current ranks as historical membership.
basis-crash-risk-research
takenthatusernameis-collab
Preregister and test whether extreme perpetual-futures basis predicts future crash frequency and downside-tail severity over 7–28 days. Use for experiment design, grid execution, outcome analysis, or interpretation; not for trading.
start-topic
kil9
새 글감(화제)을 받아 topics/ 에 기록하고, 능동적 대화로 논지를 발전시킨 뒤 반론·검증 등 보조 에이전트를 거쳐 drafts/ 에 초안을 만든다. 컨텍스트 리셋 후 첫 글 시작에도 사용한다. 완성된 콘텐츠를 페이지로 옮기기만 할 때는 /publish-til 을 쓴다.
omni-control-bridge
sparkedoficial-stack
Active bidirectional feedback and telemetry bridge with OMNI-LOCAL-TEMPORAL CONTROL (GODWORKS SYSTEM v26.4). Use to inspect system health, retrieve live self-improvement instructions, talk directly with the sovereign GIA node, and submit optimization requests.
godworks-system-suite
sparkedoficial-stack
Sovereign control over GODWORKS SYSTEM v26.4 & TARDIS. Perform web research, autonomous code evolution, query conjectures, inspect sensor/radar telemetry, and interact with the sovereign LLM directly from Orca.