unknowns

Plugin unknowns from marketplace ajitta in ajitta/claude-plugins. Commands read from the repo's manifest — 5 days ago.

Skills (11)

• blindspot

  Blind-spot pass — before work on a plan or system in any domain, investigate the unknown unknowns on both sides (what the user has not seen, what the AI does not know about their situation) and turn them into a better prompt. Use on "blind spot pass", "what am I missing", "unknown unknowns", or befo

• brainstorm

  Brainstorm the intervention — for a named problem in any domain, spread ~10 candidate interventions from a fix for this afternoon to a quarter-long bet without judging them, then size the ones the user picks. Use on "brainstorm interventions", "show me options", or when a problem is named with no ap

• buy-in

  The buy-in doc — before asking for approval of any work, surface what the approvers do not know and will object to, answer each point with evidence, state the limitations, and name who signs off. Use on "buy-in doc", "prep me for review", "get sign-off", or before requesting review, approval or merg

• deviation-log

  Deviation log (implementation notes) — while doing any planned work, log each significant departure from the plan, why, and what the user still has to judge, beside the deliverable. Use automatically when the work hits an unknown the plan did not cover, and on "implementation notes", "record a devia

• interview

  The interview — before deciding, the model asks the user, in rounds of at most four, what it cannot know or judge about their situation: open questions first, then options for each decision, biggest blast radius first; ends with a decision table and a ready prompt. Any domain. Use on "interview me",

• prototypes

  Prototypes in the design-thinking sense — several versions of anything, each on a clearly different philosophy, for the user to react to; ends with a confirmed requirements list. Use on "divergent prototypes", "design options", "versions to react to", or when the user cannot say what they want ("kno

• quiz

  Post-work comprehension quiz — explain work the model largely produced (any deliverable), then quiz the user so they can explain and defend it. Use on "quiz me", "test my understanding", or right before handing over, presenting, submitting or merging such work. Runs even when a quiz could be written

• reference

  Point at a reference — treat a provided example as a map, not an answer: read it, show the user your reading, and get it corrected before building on it. Use on "use this as a reference", "make it like this", or when the user supplies an example to build from. Runs even when the adaptation looks sim

• teach-me

  Teach me my unknowns — an interactive explainer for a field's vocabulary and mental model, in any field, so vague requests become precise. Use on "teach me", "make me an explainer", or when the user cannot name what they want or judge the result. Blindspot investigates a specific situation; teach-me

• tweakable-plan

  The tweakable plan — for any work, a plan sorted by likelihood of change, not execution order, so the decisions the user is most likely to change are reviewed first, each with alternatives. Use on "plan this", "make a plan", "tweakable plan". Not native plan mode: a reviewable document before the wo

• unknowns-loop

  Run the know-your-unknowns workflow end to end (explore → question → plan → work → verify → quiz → outcome check) on significant or ambiguous work in any domain. Use on "unknowns loop", "run the operating loop", "know your unknowns". A bare "loop" means the built-in interval runner, not this skill.

Subagents (2)

• independent-reviewer

  Independent verification agent for unknowns work — checks a finished deliverable of any kind against its plan, done criteria, outcome and deviation log WITHOUT trusting the producer's narrative: conformance, whether recorded deviations were acceptable, and cases that pass on paper but fail in realit

• unknowns-scout

  Blind-spot investigator for unknowns work — before work in unfamiliar territory, investigates the sources the task has (documents, data, the web, a codebase, its history) and surfaces unknown unknowns, risky assumptions, structural conflicts and fragile areas, then hands back an improved prompt. Spa

Similar plugins