# Skill Catalog — Domain Authority and Optimization Machinery for Case Resolution A catalog of the skills behind MongoDB and Atlas case resolution, in five families, plus the shared optimization loop and blind-backtest evidence that tie them together. **Scope note.** This catalog explains the skills across five families (**MongoDB, psychology, writing, troubleshooting, and expertise**) that together support automated and human-assisted MongoDB/Atlas case resolution. "Expertise" is covered in **both senses**: the domain-authority skills (the deep `*-expert` hubs) *and* the expertise-engineering machinery that captures, teaches, measures, and optimizes expertise. A later section explains the optimization-pipeline machinery that connects them. That machinery is also the subject of the companion whitepaper, "Compounding Expertise — An Iterative Skill-Optimization Pipeline." **How to read this.** The skill tree is organized hub-and-spoke: a **hub** skill routes to on-demand **reference files** (spokes). When an entry says "hub," it is the routing entry point; its spokes load only when a task matches. Skill names in `code font` are the exact identifiers. Reference counts in parentheses reflect the skill tree when this catalog was drafted and drift as hubs are rebuilt. --- ## 1. MongoDB & Atlas — domain authority These are the knowledge bases an answer is built from. They are split by *plane* (data vs. platform vs. operations) and by *mode* (distilled guidance vs. live diagnostics vs. raw lookup), so a task loads only the surface it needs. | Skill | What it is | Case-resolution role | | :---- | :---- | :---- | | `mongodb-expert` | Primary **data-plane and engine** hub (24 references): CRUD/MQL, aggregation pipelines and deep stages, index design, query optimization and explain plans, schema design, transactions, change streams, time-series, geospatial, BSON, error codes, driver internals (CMAP/SDAM/retryable writes), WiredTiger internals, sharding, replication. | First stop for any query/index/schema/engine question; the deepest single body of MongoDB knowledge. | | `mongodb-atlas-expert` | Broad **Atlas platform** hub (27 references): control plane, Admin API v2, Atlas CLI, Terraform, Kubernetes Operator, tiers/limits, security and private networking, alerts/metrics, plus sub-areas (Atlas Search, Vector Search, Stream Processing, Charts, Triggers, Online Archive, Federated Auth, cloud networking/PrivateLink). | Anything about the managed platform's configuration, surface choice, or security posture. | | `mongodb-operations-expert` | **Operations and reliability** hub (18 references): backup/restore and PITR, DR and RTO/RPO, Ops/Cloud Manager, upgrade paths and FCV, migration patterns (`mongosync`, Relational Migrator), CDC, security architecture, encryption (CSFLE/Queryable Encryption), compliance, cost optimization, Kafka/Spark connectors. | Reliability, data-movement, security-architecture, and cost questions: the "keep it running and safe" plane. | | `atlas-diagnostics-expert` | **Live diagnostics, performance, monitoring, and capacity** hub: internal triage tooling, FTDC (full-time diagnostic data capture) and log tooling, Performance Advisor, Atlas alert conditions, benchmarking (YCSB), observability (Prometheus/Datadog/Grafana), capacity planning. Also owns the reference for an internal diagnostics package built as a Chrome extension. | The triage surface, chosen first when a *live* case has a performance or health symptom. | | `mongodb-kb` | Index of ~2,717 **Knowledge Base articles**: error-code lookup, symptom→article matching, shareable `support.mongodb.com` URLs (Public articles only, never Internal). | Matches reported symptoms to a known issue and yields a citeable, customer-shareable reference. | | `mongodb-docset-lookup` | **Offline MongoDB Manual** lookup via a version-pinned Dash docset (3,555 pages; 6,159 indexed entries). | Grounds a generated claim in exact, authoritative Manual text; this is the fact-check substrate. | | `mongodb-university-certification` | MongoDB University platform, certifications, and enablement learning paths. | Used when the resolution is enablement (build the customer's competence), not a code fix. | | *(internal repo intelligence skill)* | Repo prioritization, symptom→repo mapping, install/run guidance, the diagnostic-tool catalog (FTDC, explain plans, Jira enrichment). | Maps a diagnostic scenario to the right internal tool or repo. | | `solve-case` | **End-to-end case solver** that orchestrates the skills above plus the case/account MCPs and the diagnostic and psychology agents: identify customer → troubleshoot/diagnose → cited analysis → psychology-informed reply + blockers/tools/escalation. | The runtime that turns intake into a drafted, defensible customer reply. | | `uber-mongodb-diagnostician` *(agent)* | Deep multi-subdomain diagnostic reasoning backed by the **66-part `uber-mongodb-skill`** monolithic reference (compiled from the family of `mongodb-*` specialist skills). Read-only; emits a rank-ordered root-cause hypothesis with diagnostic evidence to collect, remediation, confidence ratings, and citations to the exact Part(s) grounding each claim. | The productized form of the highest-scoring backtest strategy (see §6 and the companion whitepaper). | --- ## 2. Psychology — human factors A correct diagnosis still fails if the *delivery* triggers resistance, or if a human over- or under-trusts the AI's output. These skills cover the human side of resolution. | Skill | What it is | Case-resolution role | | :---- | :---- | :---- | | `applied-psychology` *(hub)* | Operator-facing applied-psychology hub, 10 evidence-based, replication-honest references: **behavior change & adoption** (Self-Determination Theory, Fogg B=MAP, habits); **decision-making** (biases, prospect theory, nudge, System 1/2); **persuasion** (ELM, dissonance, reactance, inoculation); **trust, rapport & psychological safety** (Mayer's ability-benevolence-integrity model, Trust Equation, Edmondson); **human-AI interaction** (automation bias, algorithm aversion/appreciation, calibrated reliance); **learning & expertise** (cognitive load, retrieval practice, spacing, deliberate practice); **emotion & affect** (emotion regulation, EI); **performance & resilience** (growth mindset, grit, flow, self-efficacy, burnout); **personality** (Big Five, HEXACO, Dark Triad); **moral psychology**. | The two most case-relevant spokes are *human-AI interaction*, so a technical account manager (TAM) calibrates trust in an AI hypothesis rather than blindly accepting or rejecting it, and *trust/persuasion*, so a reply repairs trust and avoids reactance. | | `customer-comms-psychologist` *(agent)* | Drafts a customer-facing communication **and** pressure-tests it against behavioral science before a human sends it. Four jobs: post-incident trust-repair, stalled-adoption nudges, renewal/expansion framing, enablement. Routes through `applied-psychology`, then `document-critique` and `kill-the-AI-ism`. | Converts a technical resolution into a psychologically sound message; the TAM owns final send. | | `psychology-of-charitable-giving`, `fundraising-and-donor-psychology`, `volunteer-and-prosocial-motivation`, `health-behavior-change-and-donor-registration`, `social-marketing-and-cbsm`, `effective-altruism-and-philanthropic-decision` | The **nonprofit/prosocial spokes** of the psychology family: donor motivation, behavior-change campaigns, volunteer motivation, effective-giving decision-making. | Out of scope for support cases; included for completeness of the psychology family. The transferable cores (behavior change, framing) live in `applied-psychology`. | --- ## 3. Writing — communication and synthesis A resolution is only as good as the artifact that carries it. This family covers drafting and, critically for the pipeline, the **critique loops** that bring an artifact to publication quality. | Skill | What it is | Case-resolution role | | :---- | :---- | :---- | | `writing-expert` *(hub)* | General prose craft, voice, style, and editing (18 sub-skills): sentence/paragraph craft; frameworks (BLUF, Minto Pyramid, SCQA, STAR, Inverted Pyramid); tone by audience; data storytelling; plain/inclusive language. | The base layer of voice and clarity for every written output. | | `technical-writing-craft` *(hub)* | Technical and product writing: API docs, READMEs, how-tos, KB articles, RFCs/design docs, **runbooks**, **postmortems**, **incident/status-page comms**, changelogs, error messages. Style-guide grounded (Google, Microsoft WSG). | The genre conventions for the operational artifacts a case produces. | | `executive-comms` *(hub)* | Leadership and persuasion artifacts: BLUF, board memos, decision memos, status updates, decks, **whitepapers**, **case studies**, OKRs, proposals. | Turns a resolution into an exec-ready readout or escalation. | | `content-and-marketing-writing` *(hub)* | External communications, including **customer-support and TAM ticket replies and escalation handoffs**, NPS/CSAT/review-response writing at scale. | The genre home for the customer-facing reply itself. | | `career-and-formal-writing` *(hub)* | Career, academic, legal, and policy/governance writing; survey-question design. | Rarely on the case path; here for family completeness. | | `kill-the-AI-ism` | Diagnostic skill that detects and replaces generator artifacts ("AI-isms") across four pattern categories, producing a findings report with human-voice replacements and an H1–H7 heuristic score. | The voice gate: strips machine tells before a reply reaches a customer. | | `document-critique` | **Multipass document review agent** (passes 0–14 plus sub-passes) with a convergence loop until no medium-or-higher findings remain; includes authoritative verification and an adversarial/hallucination guard. | The general-purpose document optimization loop; one instance of the shared machinery in §6. | | `ddo` *(Document Deep Optimizer)* | The apply-in-place driver over `document-critique`: runs the multipass critique and **applies every medium-or-higher fix**, looping to convergence (max 3 iterations, raised to 5 if Medium+ findings dropped ≥50% in the prior iteration). Fast paths: `--voice-only`, `--read-only`, `--annotate`. | One command to take a draft to convergence. | | `draft-review-revise-loop` | Meta-skill defining the explicit three-pass workflow (draft → review → revise) with hard/soft stop conditions to prevent infinite-polish loops. | The discipline that decides *when to stop* optimizing. | --- ## 4. Troubleshooting — diagnostic reasoning Domain knowledge tells you *what is true*; diagnostic reasoning tells you *how to find the fault*. This family covers both the live act and the pedagogy and evaluation of fault-finding. | Skill | What it is | Case-resolution role | | :---- | :---- | :---- | | `atlas-diagnostics-expert` *(cross-listed from §1)* | Live MongoDB/Atlas diagnostics, performance, monitoring, and capacity. | The hands-on fault-finding surface for live cases. | | `software-engineering-patterns` *(hub)* | Language-agnostic engineering practice, including **debugging and root-cause analysis (5 Whys)**, code review (OWASP checklist), performance profiling, and **automated program repair** (fault localization, patch generation). | The general RCA discipline behind code-level case resolution. | | `teaching-troubleshooting-diagnostic-reasoning` | The **pedagogy** of fault-finding: cognitive apprenticeship, productive failure (Kapur), mental-model instruction, the novice→expert trajectory, **illness scripts**, dual-process theory, **key-feature assessment**, game-day/fire-drill as pedagogy. | Designs how humans (and, by analogy, agents) are trained to diagnose novel faults; the source of the "illness script" framing used in diagnosis. | | *(internal repo intelligence skill, cross-listed from §1)* | The diagnostic-tool catalog and symptom→repo mapping. | Points the diagnosis at the right instrument. | | `mongodb-kb` *(cross-listed from §1)* | Symptom→article matching and error-code lookup. | The fast path from a symptom to a known root cause. | | `diagnosis-methodology-backtest` | Runs a **blind, parallel, multi-agent backtest** comparing competing diagnosis methodologies against ground-truth resolutions, scoring which predicts root causes most accurately. A methodology sees only what the customer first reported, never the resolution; the predictor is never the grader; each methodology runs in its own isolated subagent. | The evaluation harness that measured the skill-knowledge strategy's accuracy (§6, with the caveats stated there). | --- ## 5. Expertise — both layers Per scope, "expertise" spans two layers: **(5a)** the domain authorities that *hold* expertise, and **(5b)** the expertise-engineering skills that *build, teach, measure, and tune* it. ### 5a. Domain authorities (the experts) The `*-expert` hubs and `uber-mongodb-diagnostician` from §1, plus the diagnostic-reasoning skills from §4. These are "expertise" in the everyday sense: the skills that know the most about a domain. They are the *inputs* the pipeline optimizes and composes. ### 5b. Expertise engineering (capture → teach → measure → optimize) This layer is what makes the domain authorities *good* and keeps them good. It is the part most people miss, and it is where the pipeline's advantage comes from. #### Capture — elicit expert knowledge | Skill | What it is | | :---- | :---- | | `cognitive-task-analysis` | Front-end knowledge-elicitation methodology (CDM, ACTA, PARI, GDTA, think-aloud, concept mapping) that surfaces the tacit decision steps experts omit. It addresses the "expert blind spot": the skill's reference reports that experts omit up to 70% of their decision steps in unaided accounts. The feedstock for any skill or training that encodes diagnostic expertise. | #### Teach — encode expertise into instruction | Skill | What it is | | :---- | :---- | | `technical-instruction` *(hub)* | Family hub routing to the instruction spokes below. | | `instructional-design-course-architecture` | Course/curriculum design (ADDIE, SAM, Dick & Carey, backward design, Gagné, Merrill); learning objectives and constructive alignment. | | `technical-training-delivery` | Hands-on labs, live coding, developer/customer academies, faded examples. | | `teaching-troubleshooting-diagnostic-reasoning` *(cross-listed from §4)* | The pedagogy of diagnostic reasoning specifically. | | `genai-education-instructional-design` | GenAI applied to education: AI tutors/ITS, AI-assisted instructional design, AI-resistant assessment. | | `human-performance-technology` | Diagnoses whether training is even the right intervention (ISPI HPT, Gilbert's behavior engineering model, Mager-Pipe) before anyone builds a course. It is the "skill gap or environment gap" filter. | #### Measure — quantify expertise and learning | Skill | What it is | | :---- | :---- | | `assessment-certification-design` | Certification/exam design and psychometrics (item writing, cut scores via Angoff/Bookmark, CTT/IRT, reliability/validity, DIF). The rigor layer for measuring competence. | | `learning-measurement-evaluation` | Training effectiveness (Kirkpatrick four levels, Phillips ROI Level 5, xAPI/SCORM, transfer climate). | | `skills-taxonomies-cbe` | Skills taxonomies and competency frameworks (O*NET, ESCO, SFIA, KSAOs) and Competency-Based Education (CBE), the substrate that competence is certified against. | #### Optimize — build and tune the skills, prompts, and code themselves (the meta layer) | Skill | What it is | | :---- | :---- | | `concept-family-explorer` | Gap-discovery layer *above* deep research: maps a subject's full conceptual family, surfaces what you're missing, scores each gap, then loops `/dr` on every viable gap until the concept tree saturates. The skill-*acquisition* engine. | | `skill-creator` | Creates new skills from scratch and measures skill performance via evals. | | `skill-optimizer` | Audits and improves a skill to production quality via a convergence-loop quality gate, then syncs it to the context hub. | | `skill-tree-architect` | Whole-tree architect that audits the hub-and-spoke taxonomy against description caps and hub balance and rebalances it. | | `prompt-deep-optimizer` | Iteratively optimizes production prompts: a **16-pass audit in 5 parallel bundles**, applies every Medium+ fix, loops to convergence; recommends a training-data optimization algorithm (APE/OPRO/MIPROv2/GEPA/…). | | `prompt-helper-optimizer` *(`/ph`, `/phe`)* | One-off/exploratory prompt improvement: interpret, critique, rewrite, and (in `/phe`) save and run. | | `prompt-engineering` | The reference body: foundational and advanced prompting techniques, structured output, injection defense, evaluation. | | `code-deep-optimizer` | Multi-stage review-and-fix optimizer for source files/repos: **18-pass audit**, applies Medium+ fixes in place, verifies via build/lint/tests, loops to convergence. | | `eval-driven-development` | The build-time discipline of evaluating LLM/agent apps: the analyze→measure→improve loop with **error analysis as the engine**, the Three Gulfs (Specification/Generalization/Comprehension), and LLM-as-judge calibration (Cohen's kappa, Krippendorff's alpha). | --- ## 6. The optimization-pipeline machinery (connective tissue) The skills above are *components*. A small set of shared mechanisms connects them into a system whose predictive accuracy can be measured. The companion whitepaper covers these mechanisms in depth; they are catalogued here because they are themselves skills and artifacts. **The shared convergence-and-severity model.** One canonical internal reference defines a **severity ladder** (Blocking/Critical → Major → Medium → Minor → Nit; fix everything Medium-and-above), **7 convergence exit conditions** (clean / no-progress / content-cycling / stable-rewrite / loop-instability / iteration-cap / budget), **iteration caps** (prompts 5; skills and documents 3, raised to 5 if Medium+ findings dropped ≥50%), and a **blind re-audit gate** (a fresh-context subagent that re-audits the final artifact with no access to the findings history before "clean" can be declared). Every optimizer below is an instance of this one loop. **The deep-optimizer family** is the same loop applied to different artifact types: - `prompt-deep-optimizer`: prompts (16 passes, 5 parallel bundles) - `code-deep-optimizer`: source code (18 passes; build/lint/test verification) - `document-critique` / `ddo`: prose (passes 0–14) - `design-deep-optimizer`: UI/visual designs (11 passes); not part of this case-resolution pipeline, listed for sibling-family completeness - `skill-optimizer`: skills themselves **The acquisition and evaluation ends:** - `concept-family-explorer`: finds and fills knowledge gaps until a concept family saturates (front of the pipeline) - `diagnosis-methodology-backtest`: blind, parallel, multi-agent evaluation against ground truth (back of the pipeline) - `eval-driven-development`: the error-analysis discipline that the backtest operationalizes - `solve-case`: the runtime orchestrator that composes the optimized components on a live case ### The evidence so far On a blind panel of 244 real support cases from one large enterprise customer (Customer A here), the skill-knowledge strategy scored as shown below. That strategy is optimized `mongodb-*` skill expertise, the input the pipeline produces. The predictor never saw the resolution, and the predictor was never the grader. | Metric | Value | Meaning | | :---- | :---- | :---- | | Raw accuracy | **72.5%** | Correct over all 244 cases (partials half-credited) | | Accuracy-on-gradable | **90.3%** | Correct over the 196 gradable cases (partials half-credited; the other 48 cases had no diagnostic content to grade) | | Defensibility | **100%** | Zero **Wrong** predictions across all gradable cases | Confusion matrix: **158 Correct, 38 Partial, 0 Wrong, 48 Unverifiable**. The skill-knowledge strategy beat both an authored-flowchart bundle and a documented-flowchart corpus; combining all three (Best-of-3, where any Correct wins) added only **7** correct cases (165 versus 158; 74.0% versus 72.5% raw accuracy). The skill-knowledge strategy alone therefore accounted for 158 of those 165 correct cases, about 96%. **Read these numbers with the author's caveats.** Most ground-truth resolutions in this panel are autoclose fallbacks, where the recorded resolution echoes the customer's first message: 188 of 244 in the author's count, with roughly 8 carrying substantive engineer narrative. The scores therefore measure plausibility against thin transcripts, not verified diagnostic accuracy. The zero-Wrong result is partly structural in the grading rules. The ranking also depends on the rubric. The calibrated rubric keeps autoclose-fallback resolutions gradable when a prediction preserves the case's domain noun-phrase, which rewards a strategy that mirrors the customer's prompt, and on a stricter 20-case seed panel the skill-knowledge strategy placed third of four. "Skill knowledge wins" is rubric-conditional, not absolute. Finally, no un-optimized baseline was run on this panel, so the scores are absolute performance, not a measured lift from the optimization pipeline; the author names that baseline as future work. *Source:* the author's internal evaluation writeup and comparative review for this panel. --- ## 7. How the families compose for case resolution A live case flows through the families in order: 1. **Troubleshooting** chooses the diagnostic surface and gathers evidence (`atlas-diagnostics-expert`, `software-engineering-patterns`). 2. **MongoDB domain authority** generates and ranks root-cause hypotheses with citations (`uber-mongodb-diagnostician` over the 66-part reference; `mongodb-kb` and `mongodb-docset-lookup` for grounding). 3. **Writing** turns the cited analysis into a customer reply and an internal readout (`content-and-marketing-writing`, `technical-writing-craft`, `document-critique`/`ddo`, `kill-the-AI-ism`). 4. **Psychology** ensures the reply repairs trust and avoids reactance, and that humans calibrate their reliance on the AI hypothesis (`applied-psychology`, `customer-comms-psychologist`). 5. **Expertise engineering** runs underneath all of it: `concept-family-explorer` finds missing knowledge, the deep-optimizers tune every skill/prompt/document to production quality, and `diagnosis-methodology-backtest` measures how well the resulting diagnosis strategy predicts root causes, feeding errors back into the next optimization cycle. `solve-case` is the orchestrator that runs steps 1–4 on a single case; the expertise-engineering layer (step 5) is the offline loop that makes steps 1–4 progressively better. The companion whitepaper argues that this composition (five complementary skillsets, each driven to convergence by one shared optimization loop and validated by one blind backtest) is what produces the measured accuracy above. No baseline run isolates that effect (see the caveats in §6).