Mechanism, not vibes
Every output names the peer-reviewed framework behind it (author and year) and it is enforced before a pack ships. 78 frameworks, applied automatically, not listed on request.
78 behavioral frameworks
FeaturesNot a prompt pack.
Not a prompt pack.
A specialist for every job.
Each one applies named frameworks, runs governed workflows, and tells you how confident it is. Below: the moat, the persona commands and what they actually do, and the enterprise-architecture layer most tools never ship.
The moat, in plain terms
Six things a generic LLM wrapper cannot do, applied to every output automatically.
Confidence you can audit
Every output carries 🟢 / 🟡 / 🔴 with a four-line rubric: Basis, Gap, Mitigation, Next. You see exactly what the model is sure of and what it is not.
Confidence rubric on every output
Compliance that blocks, not advises
14 regulated frameworks wired at the MCP layer. An output touching PII, PHI, or PCI without the right control is stopped at generation, not flagged in a review after it shipped.
14 frameworks · enforced at the gate
Drift caught before it ships
13 drift classes scan on every session start: docs vs code, counts vs filesystem, even your own marketing claims vs the moat ledger. Silent regressions surface before a customer or auditor finds them.
13 drift classes · every session
Autonomy that can't run away
Unattended workflows with a budget guard that hard-kills on token, tool-call, runtime, or cost breach, a verification gate that runs separately from generation, and dry-run on by default.
Budget guard · verification gate · dry-run
Governance by default
Every prompt routes to the right specialist automatically. You never memorize a command. The behavioral layer engages whether or not you ask for it.
Routes every prompt · Claude Code CLI
One command, the right specialist
Speak your role's vocabulary; Maxim routes invisibly to the specialist who owns it, loads the frameworks below, and can run the workflow unattended.
/mxm-founderEarly-stage founder
Positioning, pricing, and a pitch that survives a skeptical investor. Out-position a competitor from a teardown, price against real willingness-to-pay, and structure the narrative so the ask lands.
Duarte SparklineMinto PyramidAARRRProspect TheoryVan WestendorpStrategyzer BMC
Workflow: competitor-monitor pings you only on a real move, with the teardown already written (dry-run first).
/mxm-archEnterprise architect
Target-state architectures, C4 model diagrams, ArchiMate views, Wardley maps, technology radars, and the ADRs authored for you. Maxim ran this discipline on itself: 21 ratified decision records.
TOGAFC4 ModelArchiMateWardley MappingADR authoringTech Radar
Workflow: architecture-drift watch flags when the codebase diverges from the documented target architecture.
/mxm-legalIn-house counsel / GRC
Jurisdiction-aware review across 14 compliance frameworks with the exact clause cited, not a summary that reads well and proves nothing.
GDPRPIPEDAHIPAAPCI-DSSSOC 2EU AI Act+8 more
Workflow: ROPA + DPIA drafting under the CSO compliance gate; sources scanned for PII before anything leaves the session.
/mxm-secureCISO / AppSec
Threat models and security reviews with triple-OWASP coverage, structured threat-modeling, AI-specific risk, and a software + AI bill of materials generated, not hand-waved.
OWASP Top 10 ×3STRIDE / PASTA / LINDDUNNIST CSF + AI RMFMITRE ATLASSBOM / AIBOM
Workflow: overnight security sweep opens a PR with findings for your morning review, never merges itself.
/mxm-pmProduct manager
PRDs, prioritization, and roadmaps grounded in real method instead of opinion: outcomes, not feature lists.
Jobs-to-be-Done (Jobs Atlas)OKR (leading + lagging)RICEINVEST
Workflow: user-feedback sweep clusters raw signals into JTBD outcomes and surfaces the highest-RICE next bet.
Enterprise Architecture & Data GovernanceMost tools cite an ADR.
Most tools cite an ADR.
Maxim writes them for you.
The enterprise-architect and data-architect specialists turn intent into the artifacts a real architecture function produces, and Maxim proved it on itself, authoring its own 21 architecture decision records.
Architecture
TOGAF target architectures, C4 model diagrams, ArchiMate views, Wardley maps for strategic positioning, technology radars, and ADRs authored with the mechanism, the anti-pattern, and the retirement condition spelled out.
TOGAFC4 ModelArchiMateWardley MappingTech RadarADR authoring
Data Governance
DMBOK across its knowledge areas (data governance, master & reference data, document & content, warehousing, data quality) plus Zachman for enterprise taxonomy. The data-architect specialist applies the framework, not a generic best-practices list.
DMBOK (DAMA)ZachmanMaster & Reference DataData QualityData Architecture
The proof is this product. Every positioning claim Maxim makes traces to a public, ratified ADR: the same discipline the architect persona gives your project. Read them on the Why Maxim page.
Skills, loops, and workflows
A skill runs once. A loop repeats until it's right while you watch. A workflow runs unattended, with brakes. Plain-English verbs reach all three.
Build a feature with TDD discipline + a Fogg B=MAP scope check, CSO auto-loop on regulated data.
Fix a bug with Systematic Debugging + root-cause discipline, tester and reviewer in the loop.
Cut a release: CHANGELOG, SBOM check, reviewer + CSO + CMO coordinated, version bump.
Plan a sprint with a Coverage Matrix + Fogg B=MAP scope gate over the planning pack.
Review code or a doc with a framework-citation requirement and a security auto-loop.
Author and run an unattended workflow under the Autonomous Workflow Standard: budget-guarded, dry-run first.
Install free.
Upgrade when the moat shows up in your work.
90-day Pro Trial auto-activates on install. Keep using Starter after, or commit to Solo at $19.99.