mbp

Plugin mbp from marketplace pfeff in pfeff/claude-skills. Commands read from the repo's manifest — 2 days ago.

Skills (42)

• cadence-goals

  Manage daily, weekly, monthly, quarterly, and yearly goal-setting loops from Obsidian periodic notes. Use when defining, reviewing, aligning, or updating cadence goals, when distinguishing today's execution from tomorrow or this week/month, or when converting a domain loop into nested annual, quarte

• ci-feedback-loop

  Monitor PR check status after push, auto-fix CI failures, and escalate when unable to fix. Use when an agent has pushed commits to a PR and needs to ensure checks pass before continuing.

• compound

  Document a solved problem as a searchable solution note in the host's Obsidian vault. Captures a single fix with YAML frontmatter for QMD-based agent discovery; delegates the file write to the obsidian-notes skill.

• decision-rights-doctrine

  Shared doctrine reference for which decisions an agent MUST escalate to the operator vs may decide autonomously. Holds the two-tier escalation model (Tier 1 hard-escalate, Tier 2 self-check-then-escalate-if-material), the six-category taxonomy (A-F) mined from guardian PRs, the alignment-audit self-

• diataxis

  Project documentation using the Diataxis methodology. Scaffold docs structure, write documentation by type, or audit existing docs. Use when the user wants to create, organize, or improve documentation.

• dispatch-gate

  Dispatch-readiness gate for background jobs. Before the operator launches a background agent (worktree job, subagent, or cross-repo dispatch), this skill checks the privacy gate (public destination + sensitive context?), classifies the job's billing pool (subscription vs metered — e.g. claude -p/Age

• drift-check

  Report-only health check that mechanically detects spec-vs-reality drift in the skills estate: merged-but-unreleased plugin content, dangling file references in skill docs, and registry mismatches between skills/ dirs and marketplace.json. Operator-invoked via /drift-check. Findings only — never edi

• family-check

  Define the guardian Family check by operator interview: elicit an observable Good / Baseline / Improvement triple for ONE family member × sub-domain at a time, then land it in the ratified source-of-truth file (mission/family-objectives.md) via branch + PR, merge-gated on the operator. Fires on "def

• fix-then-re-review-ladder

  Recovery loop driving a PR (or reviewed work product) from a NEEDS-WORK/findings-attached verdict back to CLEAN. Triggers when a /review, /l1-review, or /l2-review verdict is not CLEAN and the operator wants it resolved. Triages findings into fix-now vs operator-waivable, dispatches ONE scoped fix a

• git

  Git usage patterns and conventions. Use when the user requests commits or git operations. Enforces compact commit messages (headline + 1-2 sentences), conventional commit format, no Claude annotations, and focuses on WHY not WHAT in summaries.

• github-projects

  Manage GitHub Projects V2 boards — list items by sprint, set status/sprint fields, add issues to boards, query field metadata. Use when working with GitHub project boards, sprint assignments, or board status transitions.

• goal-tree

  Orchestrate multi-task projects as hierarchical goal trees. Use when a project spans multiple tasks, repos, or sessions and needs coordinated decomposition, parallel dispatch, and synthesis. Sits above task-workflow — feeds work into workspaces, does not replace them.

• handoff

  Write and rehydrate a durable session-handoff document so a fresh, zero-context session resumes work without inheriting bloated history. Use when approaching the context limit, before /clear, at end of session, or when handing work to another session or agent. Two operations — write (capture current

• improvement-loop

  Adopt the improvement-loop role — the 'meta' track that runs alongside the operator's mission work (the operator's master session, advancing work via background subagents/workflows), either as the meta pane of a two-pane working session pair or as a standalone CoS-owned session. Use when told 'you a

• kb-capture

  Capture / ingest bridge for the LLM Knowledge Base. Sweeps Readwise HIGHLIGHTED sources (the operator highlights-then-archives, so highlights — not the inbox — are the signal) into the vault's raw/ queue for later /kb-compile; kb-tagged docs are captured unconditionally. A source is captured iff it

• kb-compile

  Compile staged knowledge-base sources into the vault's native notes: one per-source summary (type: reference) plus fuller concept notes (type: zettel) in Notes/, cross-linked to Keywords/ term pages and surfaced through a KB MOC + Dataview. Incremental and idempotent; concept notes update in place.

• kb-core

  Shared library for the LLM Knowledge Base skills (kb-capture, kb-compile, kb-lint). Holds the SPEC v2 §1 primitives — capture predicate, idempotency key, off-limits write-guard, KB-ownership check, concept-slug dedup, orphan detection — in one place so the three skills share one implementation. Refe

• kb-lessons-bridge

  Bridges reading-derived workflow insights into the self-improvement loop. Reads a kb-compile output note (project:knowledge-base, type:reference/zettel) or a Knowledge Intake synthesis note's "## Workflow implications" / "## Candidate actions" sections, filters for process/workflow-relevant insights

• kb-lint

  Report-only health check over the LLM Knowledge Base notes (reading the wider vault for connection candidates). Surfaces inconsistent/contradictory data, orphan notes, missing data web search could impute, and candidate new connections. Findings are report-only — any auto-fix is confined to KB-owned

• l1-supervisor

  Adopt the L1 supervisor role for a goal tree. Use when a session is told "you are L1", when /mbp:l1-supervisor is invoked, or when the operator describes L1 supervision. Installs the standing contract — identity, source-of-truth precedence, tick procedure, permission and escalation rubrics, stop sig

• l2-supervisor

  Adopt the L2 supervisor role for a goal tree. Use when a session is told "you are L2", when /mbp:l2-supervisor is invoked, or when the operator describes L2 supervision. L2 is a KPI-driven controller that keeps three axes — Progress, Budget, Health — in band by proposing intervention L1s to the oper

• lN-lifecycle-doctrine

  Shared doctrine reference for child-session lifecycle management at every supervision layer. Defines the generic child state machine (spawned→working→complete→retired, context-pressure axis, edge states), per-state entry/act/exit rules, complete-check guards, teardown discipline, idle-fleet handling

• lN-review-doctrine

  Shared doctrine reference for the Acceptance and Objective review skills (l1-review, l2-review). Holds the 3-axis checklist (Conformance, Process, Objective Advancement), the reviewer-independence and reviewer-worktree-discipline invariants, the work-type → required-verification map consumed by axis

• lessons-learned

  Interactive retrospective discussion about session problems, generating actionable recommendations

• medical-records

  File a medical After Visit Summary into the Obsidian vault as a structured, queryable record. Extracts vitals and visit metadata from the document text, writes a dated note with flat Number frontmatter properties (so Obsidian Bases and Dataview can trend vitals across visits), archives and embeds th

• operator-interview-doctrine

  Shared doctrine reference for operator-interview skills. Holds the question taxonomy (context-free questions first, then funnel), the ask-vs-default rule gated on reversibility, the adaptive-depth rubric, stop conditions, AskUserQuestion batch rules, the spec-note schema, and the sign-off protocol.

• operator-interview

  Interview an operator to elicit a buildable specification, then write it as a signed-off spec note in the host's Obsidian vault. Runs a hybrid funnel — context-free open questions, AskUserQuestion scoping batches with recommended defaults, gap probes, a bounded clarification gate, playback, and expl

• operator-review

  Present a completed change to the human operator to prove it solves the problem and obtain sign-off, in a working session. Fires when work is ready for the operator's decision — "get my sign-off", "present it for approval", "present it in chunks", "use the interview tool", "review these with me". Dr

• planning-workflow

  Plan implementation tasks by validating the problem, reconciling DESIGN.md, searching past solutions, calibrating research depth, analyzing edge cases via SpecFlow, and generating living plans with checkable criteria. Use when starting implementation planning for a task or issue.

• pr-review-fanout

  Run the L1/L2 agent-review ladder across multiple open PRs concurrently instead of serially. Triggers when 2+ open PRs each need /l1-review and /l2-review before human review. Dispatches one delegated review agent per PR (each runs /l1-review then /l2-review sequentially, reading the posted L0 /revi

• pull-task

  Pull the next task from a GitHub project board or Jira. Queries project items, filters by status and sprint, presents ranked candidates, and outputs the selected issue for workspace creation. Use when the user needs to pick their next task from the project board.

• ralph-wiggum

  Scaffold a project for the Ralph Wiggum autonomous control loop. Use when setting up a new project for autonomous iteration, creating loop configuration files, or preparing specs for AI-driven development. Supports single-repo and multi-repo workspaces.

• review

  Spawn parallel specialist review agents (security, simplicity, architecture, correctness) against a PR or branch diff. Synthesizes and publishes findings; owns authorized worker handoff and remediation through independent re-review. Use for code review, review handoff, or monitoring review fixes.

• root-cause-depth

  Use before declaring a root cause "found" in any diagnosis, debugging, incident, or postmortem task — especially when you've identified what is wrong and are about to propose a fix. Runs an evaluable stop-condition test that pushes past the proximate symptom to a named, owned mechanism. Triggers: "r

• scan-document

  Capture a scanned document (PDF or image) into text, classify what it is, and route it to the right follow-up workflow. This is the front door for any scan the user drops in — after-visit summaries, lab results, receipts, letters, contracts, statements. OCRs natively on macOS (Vision framework as th

• self-improvement

  Read recommendations from lessons learned documents and incrementally improve skills, commands, and agents. Use when user asks to improve skills from lessons, apply recommendations, or requests "/claude-skills:self-improvement".

• self-verify

  Self-verification capability for dispatched jobs. Before reporting "done," a job invokes this skill to check its own change against the 3-axis review doctrine (Conformance / Process / Objective-Advancement) by composing existing review tooling and tests, then emits a structured annotation artifact.

• skillify

  Fast-capture entrypoint for turning a workflow just performed in this conversation into a skill. Reads back over the recent turns, drafts a lite SKILL.md (trigger description + steps + edge cases), and confirms with the user before writing anything. Hands off to skill-creator for the full interview/

• task-workflow

  Manage development task workspaces with standardized structure including workspace creation, git worktree management, tmux session orchestration, GitHub issue integration, and progress tracking. Use when creating tasks, setting up workspaces, resuming work, listing tasks by status, navigating task p

• testing-without-mocks

  Test-Driven Development (TDD) using the Testing Without Mocks pattern language. Create fast, maintainable, refactorable tests using Nullables instead of mocking frameworks. Supports sociable, state-based tests without broad integration tests. Use when implementing features with TDD, testing infrastr

• vault-cloud

  Record and search the Obsidian vault from a cloud session (Claude Code on the web), where the macOS-local obsidian CLI is unavailable but the vault is reachable as a git repo. Clones the configured vault repo (OBSIDIAN_VAULT_REPO), searches it with ripgrep (lexical — QMD/semantic search stays on the

• verify-inherited-premise

  Use when dispatching a task (or about to act) whose premise comes from an inherited claim rather than direct observation of the live system — a handoff/HANDOFF.md doc, an inbox or spec artifact, a skill's own documented schema/paths/behavior, a prior agent's summary, or a PR-body assertion. Adds a f

Similar plugins