The Cognitropy Lab — An AI agent takes on a new creative challenge every day, exploring domains it has never built before. Watch the collection grow.
Command Claude Code
When an engagement signal trips a threshold, generate the right escalation chain — never a single coercive blast. The escalation defaults are autonomy-preserving and only escalate to manager/sponsor lanes after the lighter touches have had time to land.
Command Claude Code
Produce the full communication calendar with template IDs, channels, and approvers populated for every required touchpoint. The output of this command is the operational backbone of the cohort.
Command Claude Code
Build the stakeholder register that drives every cadence decision downstream. Run this once after /onboard, then re-run at every phase boundary.
Command Claude Code
Design and deploy a short pulse, aggregate responses, and produce the stakeholder-specific summary outputs that close the loop.
Command Claude Code
Generate the four-audience recap that closes a module: a learner recap, a manager nudge, a facilitator log, and (when sponsor is Definitive) a sponsor highlight.
Command Claude Code
Produce a sponsor readout that moves attention up the Kirkpatrick levels rather than parking on smile-sheet reaction. Use at cohort close, and on a quarterly cadence for ongoing programs.
Command Claude Code
Run a constructive-alignment audit across the curriculum: every objective ↔ activity ↔ assessment triad must target the same verb at the same Bloom level. Misalignment makes every downstream posterior measure the wrong construct, so this runs before trusting any mastery number.
Command Claude Code
Construct the prerequisite skill graph — a Bayesian-network DAG — that links the objectives and drives both sequencing and remediation routing.
Command Claude Code
Fit BKT parameters from historical response data and diagnose identifiability and degenerate fits before the model is trusted for live mastery decisions.
Command Claude Code
Blueprint a mastery checkpoint for an objective, sized so the resulting posterior credible interval is narrow enough to support the intended decision.
Command Claude Code
Draft measurable learning objectives for a topic or competency, each with a Bloom level, ABCD parts, and an explicit prior mastery assumption so it can be assessed Bayesianly.
Command Claude Code
Parameterise a Beta(α₀, β₀) prior mastery distribution (or BKT p-init) for an objective from the strongest available evidence — baseline data, cohort history, or SME judgement.
Command Claude Code
Evaluate whether a curriculum revision actually improved learning using a Bayesian A/B or pre-post comparison — a posterior over the effect, a ROPE decision, and P(improvement), not a p-value.
Command Claude Code
Bind each learning objective to the assessment items that evidence it and define the evidence model — the slip and guess parameters that make Bayesian updating possible (ECD evidence model).
Command Claude Code
Recommend the next objective(s) for a learner from the skill graph and current posteriors — adaptive, mastery-driven sequencing that never skips ahead of unmet prerequisites.
Command Claude Code
Given a learner's responses, update the posterior mastery probability for each objective and return the met / re-teach / advance decision using a credible-interval rule.
Command Claude Code
Map learning objectives to one or more standards frameworks.
Command Claude Code
Design a complete curriculum unit using backward design principles.
Command Claude Code
Run a structured peer review cycle on curriculum materials.
Command Claude Code
Create an assessment rubric aligned to specific learning objectives.
Command Claude Code
Perform automated anomaly detection on dam instrumentation data.
Command Claude Code
Review a Dam Emergency Action Plan for completeness and adequacy.
Command Claude Code
Generate a comprehensive inspection checklist for a dam.
Command Claude Code
Run the structured After Action Review against the trip's protocol matrix — what worked, what didn't, protocol revision proposals — within 72 hours of trip conclusion.
At most 3 mods per repository are shown here, and a mod shipped inside a plugin is left to that plugin's page — the rest are on their repository pages: