Business & Operations

Core modeling methodology for TLA+ state machines. Use when choosing variables defining Init and Next designing actions structuring the state space or deciding abstraction level for concurrent or distributed systems. Trigger on questions about how to model a system as a state machine or how to write good actions and Next relations.

Systematic method for extracting faithful TLA+ specifications from real source code in Rust Go Erlang Python or other languages. Use when the user supplies implementation code and wants a matching formal model or when checking conformance between code and an existing spec. Trigger on code-to-spec reverse engineering or implementation-aware modeling.

Stage 4 tab (TLA+) of the Mermate pipeline — formal spec generation from a mastered run, SANY parse, TLC model check, spec edit/revalidate, and the Specula engine bridge. Use when editing TLA+ generation, verification, repair, the /api/render/tla endpoints, or specula-engine integration. Formal-method decisions defer to the specification-master-agent tree.

Stage 2 tab (Markdown Spec) of the Mermate pipeline — markdown architecture spec intake (paste/upload .md/.markdown/.txt), agent planning and spec refinement, and the Markdown → Mermaid unlock. Use when editing md-stage UI, upload handling, or spec refinement behavior.

OODA loop decision framework (Observe, Orient, Decide, Act). Use for complex decisions, problem-solving, unclear situations, or when someone is jumping to solutions without analysis.

Universal context persistence across sessions. Auto-saves decisions, conventions, and progress to .rune/ files. Loads state at session start. Use when any skill makes architectural decisions or establishes patterns that must survive session boundaries.

Engineering retrospective. Analyzes commit history, work patterns, and code quality metrics with trend tracking. Per-person breakdowns, shipping streaks, and actionable improvements. Use when asked for \"retro\", \"weekly review\", \"what did we ship\", or \"engineering retrospective\".

Structured reasoning frameworks for complex problems. 19 analytical frameworks, 12 cognitive bias detectors, 10 decomposition methods, 10 mental models, Cynefin domain classification, ethical dimension check, and 6 communication patterns. McKinsey-grade problem solving for AI coding assistants.

Create structured implementation plans from requirements. Produces master plan + phase files for enterprise-scale project management. Master plan = overview (<80 lines). Phase files = execution detail (<200 lines each). Each session handles 1 phase. Uses opus for deep reasoning.

Cross-session cognitive persistence via Neural Memory MCP. Use when needing semantic recall of past decisions / errors / insights across projects — distinct from session-bridge (file-based, project-scoped). Provides hypothesis tracking, evidence chains, and graph-based associative memory.

Persistent state tracking and Architecture Decision Records across sessions. Use when recording a decision, ADR, or progress that must survive session boundaries. Manages progress state, module health, dependency graphs, and ADRs for any workflow.

Find architectural friction in a codebase and propose deepening opportunities. Use when user wants to improve architecture, find refactor candidates, consolidate shallow modules, or make a codebase more testable. Outputs scored proposals (depth/leverage/locality) that surgeon and review can consume.

Generate and parse office documents — PDF, DOCX, XLSX, PPTX, CSV. Use when creating reports, exporting tabular data, or processing uploaded office files. NOT for project documentation (use docs).

Use when a reader needs an editorial architecture, flowchart, sequence, state, ER, swimlane, loop, layer-stack, data-flow, or process diagram as self-contained HTML/SVG. Symptoms: 'draw the system', 'architecture diagram', 'sequence of calls', 'state machine visual'. Not for logos, OG images, or Marp decks.

Dependency health management. Detects package manager, checks outdated packages and vulnerabilities, and produces a prioritized update plan.

Context window management. Auto-triggered when context is filling up. Triggers smart compaction and preserves critical information across compaction boundaries. Called by L1 orchestrators at context thresholds.

Creative ideation and solution exploration. Generates multiple approaches with trade-offs, uses structured frameworks (SCAMPER, First Principles), and hands off to plan for structuring.

Stage 4 (final) of the WinDAGs meta-DAG. Assembles the final PredictedDAG from Sensemaker, Decomposer, Skill Selector, and PreMortem outputs. Picks planning topology, names runtime divergence honestly, validates against the schema, and emits a runnable plan. Activate on "synthesize predicted DAG", "assemble final plan", "topology selection", "runtime honesty", "predict assembly". NOT for generating subtasks (windags-decomposer), classifying problems (windags-sensemaker), picking skills (winda...

State management, checkpoint architecture, and recovery strategy selection for WinDAGs V3 execution infrastructure. Teaches the validated state machine (transition matrix with fail-fast on illegal transitions), checkpoint-resume protocol (pluggable storage, auto-save, selective re-execution), recovery strategy selection (6 strategies classified by error type and escalated by cost), and why idempotent re-execution beats saga compensation for LLM agent orchestration. Activate on "resilience", "...

Failure pattern scanning and risk assessment for WinDAGs DAGs. Runs after decomposition but before execution. Performs a lightweight scan on EVERY DAG (BC-PLAN-004), escalating to deep analysis when recognition confidence is low or known failure patterns are found. Activate on "premortem", "failure scan", "risk assessment", "DAG validation", "timing analysis", "failure patterns", "pre-execution check". NOT for post-execution learning (use windags-curator), retrospective analysis (use windags-...

Polya's four questions after every WinDAGs execution. Q1 (contract satisfied?) and Q2 (unstated assumptions?) are mandatory on every DAG. Q3 (generalizable?) and Q4 (broader connections?) are conditional on quality and cost thresholds, run asynchronously, and never delay result delivery. Activate on "looking back", "Polya", "retrospective", "post-execution review", "was the contract satisfied", "unstated assumptions", "generalize", "problem class". NOT for learning engine updates (use windags...

Maintain consistency across all winDAGs documentation surfaces — corpus documents, marketing copy, README files, site content, skill files, API docs, and derivative documents. Enforces the canonical terminology table, voice guidelines, architectural accuracy, and cross-reference integrity. Activate on "windags docs", "documentation consistency", "windags copy", "terminology check", "docs audit", "windags readme", "site copy", "docs sync", "content review", "windags librarian". NOT for writing...

Post-execution skill crystallization and learning engine updates for WinDAGs. Runs after successful execution to update Thompson sampling parameters, track method quality, detect monster-barring, log near-miss events, and signal Kuhnian crises. Activate on "curator", "learning update", "skill crystallization", "Thompson sampling", "monster-barring", "near-miss", "Kuhnian crisis", "post-execution learning". NOT for pre-execution risk scanning (use windags-premortem), retrospective analysis (us...

The living embodiment of WinDAGs V3 — the AI orchestration platform where agents accumulate genuine expertise. Knows the full constitution, convention process, tradition positions, behavioral contracts, and derivative documents. Advises on architecture with ADR provenance, explains decisions with tradition attribution, grows source material as we build. Activate on "windags", "WinDAGs", "constitution", "convention", "why did we decide", "what does WinDAGs say", "how should we build", "archite...