Common Lab（實驗性）

Plugin common-lab from marketplace common-dev in TLOGBen/common-dev-plugin. Commands read from the repo's manifest — 25 days ago.

Skills (15)

• cold-estimation-jever

  analysis-estimation:cold-estimation 的 Lab 疊加版：在工項清單與定價之間加上 TypeSafe Jev 的判斷——每個工項的困難度分數（決定哪些要先做窄技術覆核）、必要性檢查（抓範圍擴張）、定價後每列人天的依據可信度（決定要向定價者追問哪幾列）。使用者要求 cold-estimation-jever 或「用 Jev 輔助初估」時使用；一般初估用 cold-estimation。

• define-goal-jever

  Lab variant of define-goal that adds TypeSafe Jev checks at its judgment points — a one-request quality check of the draft goal before pinning, a pre-question check before asking the user, a pinned-goal comparison, and goal-type routing to the right quantification heuristic. Use when the user asks f

• init-jev

  Set up TypeSafe Jev access — make TYPESAFE_API_KEY available to the agent's shell without exposing the key, cached for 30 days by default so the password manager is not asked on every shell — then verify with one cheap call. Use when the user asks to set up Jev or a TypeSafe key, or when a jev-gate,

• is-truely-need-to-ask-user-jev

  Before you ask the user anything, ask TypeSafe Jev whether the question truly needs the user or whether you could find the answer — or do the thing — yourself. Use every time you are about to ask the user a question, request a choice, ask them to do something for you (start a service, run a command,

• jev-browser

  Browser debugging with TypeSafe Jev element picking, built on the external dev-browser CLI: index the visible interactive elements, let Jev pick the target in one Choice call, and read only the candidates instead of the whole DOM. Invoke explicitly as /common-lab:jev-browser; use the dev-browser CLI

• jev-gate

  Get a fast, calibrated yes/no probability from TypeSafe Jev (Noul) on a piece of text. Use it on your own whenever you face a yes/no judgment that a quick second opinion would settle — is this command destructive, does it leak credentials or files, is it beyond what the user asked, is this line user

• jev-pick

  Pick one option from a known list with TypeSafe Jev (Choice), with a probability for every option. Use it on your own whenever you must choose among enumerable options for a piece of text — which page element to click, which skill or tool fits a request, which category, team, or failure type somethi

• jev-score

  Grade text on an ordered rubric with TypeSafe Jev (Score), with a probability for each level. Use it on your own whenever you need a quick independent rating of text on fixed levels — how well a completion claim cites evidence, how severe a finding is, how clear an instruction or question is — as a

• jever

  Use Jev to strengthen anywhere that only needs a judgment — yes/no, choose one, score. Look at the current target (work in progress, a prompt, a skill, an MCP server, a hook, a script or agent loop) and point out where a TypeSafe Jev call could replace, speed up, or back up a judgment, or where Jev

• review-jever

  Lab overlay on baransu:review that adds a TypeSafe Jev confidence section — for each retained finding, how well the cited evidence supports its claimed consequence and whether it is a defect, an unknown, or an optional improvement — as a second opinion shown beside the review's own judgment. Use whe

• strategic-advance-jever

  Lab overlay on common:strategic-advance that adds TypeSafe Jev readings at four judgment points — tagging each intelligence item against the MOE before a scribe consolidation, a posture reading (full / light / stand-down) at the fixed self-check, split/merge confidence when regrouping fronts, and a

• suggest-jev

  Router that lets TypeSafe Jev suggest how to handle a request before any work starts — which skill to load first, which others will be needed, the workflow, how to prove it is done, whether to dispatch subagents (how many, which model family and effort), the difficulty, and the impact — then hands s

• think-jever

  Lab overlay on baransu:think that adds a TypeSafe Jev confidence section — how well this run's evidence supports each bet the stance rests on and the stance as a whole, whether a guessed phrase in the restatement would change the shape of the answer, and whether a remaining choice is genuinely the u

• wait-what-jever

  Lab overlay on common:wait-what that checks the re-explanation with TypeSafe Jev before sending it — would this audience actually understand it — and, when not, lets Jev pick among a few candidate explanation shapes before rewriting.

• wayfinder-jever

  Lab overlay on common:wayfinder that adds TypeSafe Jev readings — which grilling questions are worth asking and in what order of impact, whether a drained ticket actually needs the human, whether a question is sharp enough to be a ticket or stays fog, which open tickets a new decision pushed out of

Similar plugins