amazon-dynamodb

name	amazon-dynamodb
description	Designs, reviews, and debugs DynamoDB data layers from design axioms — enumerates access patterns, chooses partition/sort keys and GSIs, decides single-table vs. multi-table, configures Streams, Global Tables, TTL, and zero-ETL integrations to OpenSearch/Redshift/SageMaker Lakehouse, and produces a defensible data-layer design with a monthly cost estimate and optional live validation. Applies whenever a user is designing, reviewing, or refactoring anything backed by DynamoDB — schemas, access patterns, GSIs, single- vs. multi-table choices, Streams consumers, transactional outboxes, Global Tables, zero-ETL pipelines — even when they don't say "axioms" or "design review." Also applies when debugging hot partitions, throttling, unbounded Scans, LWW conflicts, or surprise bills on DynamoDB workloads.
version	1

DynamoDB Axioms

This document is a set of design axioms for DynamoDB applications. It is intended to be read by an agent with no other context about the application and used to produce a defensible data-layer design.

Resolving the skill's own paths

This skill is host-agnostic — it runs under Claude Code, Kiro, Codex, Cursor, a plain terminal, or CI. Where it lives on disk depends on the host (~/.claude/skills/…, ~/.kiro/…, ~/.codex/…, ~/.cursor/…, a repo checkout, anywhere). The agent's working directory is the user's project, not the skill bundle, so relative paths like scripts/calculate_costs.py will not resolve. Throughout this document, ${SKILL_DIR} means the absolute path of the directory that contains this SKILL.md file (the skill root, which holds scripts/ and references/).

Resolve ${SKILL_DIR} once per session, then reuse it. Pick the first method that works in your host:

1. You already know it. You loaded SKILL.md from a path — ${SKILL_DIR} is the directory that file is in. This is the most reliable source; prefer it.

2. An environment variable. If $DDB_SKILL_DIR is set, trust it.

3. The bundled resolver (host-neutral, no host assumptions). It searches the common install roots and verifies the hit against sentinel files, so it never returns the wrong directory silently:

   # If you already know the path to the script, just run it directly:
   #   SKILL_DIR="$(sh /path/to/amazon-dynamodb/scripts/find_skill_dir.sh)"
   # If you don't, this host-neutral one-liner searches the common roots
   # (~/.claude, ~/.kiro, ~/.codex, ~/.cursor, ~/.config, ~/.local/share, $PWD):
   SKILL_DIR="$(find "$HOME" "$PWD" -maxdepth 7 -type f -name SKILL.md -path '*amazon-dynamodb*' 2>/dev/null \
                | head -1 | xargs -I{} dirname {})"
   # Verify it before trusting it (sentinel check), then hand off to the resolver
   # for its loud-on-failure diagnostics:
   SKILL_DIR="$(sh "$SKILL_DIR/scripts/find_skill_dir.sh" 2>/dev/null || echo "$SKILL_DIR")"
   

   The resolver prints the verified skill root and exits 0, or prints nothing and exits non-zero with a fix-it message — so SKILL_DIR="$(sh …/find_skill_dir.sh)" is safe to trust when it succeeds. It is plain POSIX sh, so it behaves identically across hosts.

Once resolved, export it so every later command is a clean substitution and the scripts can also pick it up:

export DDB_SKILL_DIR="$SKILL_DIR"
python3 "$DDB_SKILL_DIR/scripts/calculate_costs.py" --model dynamodb_data_model.json --output cost_report.md

Internally the scripts locate their own siblings (other scripts, scripts/benchmark_lambda.py) relative to themselves, so you only ever need the root path — never each individual script path.

Rules:

• Always invoke scripts with an absolute path (the $DDB_SKILL_DIR/… form). Do not cd into the skill directory — the user's working directory must stay put so their artifacts (dynamodb_data_model.json, cost_report.md, …) land where they expect.
• If none of the three methods resolves the directory, stop and ask the user where the skill is installed rather than guessing. A wrong ${SKILL_DIR} produces confusing "file not found" failures downstream; one clarifying question is cheaper.

The pipeline at a glance

The skill is one tool per stage. The default path touches no AWS: most work is stage 1 (a design you can discuss and refine conversationally). Stage 2 (cost) runs on request or when the design is being finalized — not reflexively every turn. Stages 3–6 are a distinctly opt-in, heavyweight fork that creates real AWS resources and incurs a real bill; enter it only on explicit user agreement. Each stage's detailed contract is in the section named in the last column.

#	Stage	Command (after export DDB_SKILL_DIR=…)	Reads	Writes	AWS?	Section
1	Design	(no script — you produce the access-pattern list + schema)	—	(in-reply artifacts)	no	Artifacts to produce
2	Cost	python3 "$DDB_SKILL_DIR/scripts/calculate_costs.py" --model dynamodb_data_model.json --output cost_report.md	dynamodb_data_model.json	cost_report.md	no	Cost estimation
3	Deploy	python3 "$DDB_SKILL_DIR/scripts/deploy_model.py" --model dynamodb_data_model.json --config benchmark_config.json --manifest-out created_resources.json --yes-deploy	model + config	created_resources.json	yes	Live validation
4	Benchmark	python3 "$DDB_SKILL_DIR/scripts/benchmark_model.py" --model dynamodb_data_model.json --config benchmark_config.json --manifest created_resources.json --raw-out perf_raw.jsonl --summary-out perf_summary.json	model + config + manifest	perf_raw.jsonl, perf_summary.json	yes	Live validation
5	Report	python3 "$DDB_SKILL_DIR/scripts/generate_perf_report.py" --model dynamodb_data_model.json --summary perf_summary.json --output performance_report.md	model + summary	performance_report.md, design_findings.json	no	Live validation
6	Teardown	python3 "$DDB_SKILL_DIR/scripts/generate_teardown.py" --manifest created_resources.json --out teardown.sh → review → bash teardown.sh --confirm	manifest	teardown.sh	yes (on --confirm)	Live validation step 6
—

Who reads what. You (the agent) read the compact artifacts: cost_report.md, design_findings.json, loop_state.json. The user reads performance_report.md. Never read perf_raw.jsonl (large) — it only feeds stage 5.

Consent gates. Stage 3+ needs --yes-deploy; the benchmark refuses to spend over cost_guardrail_usd without --allow-spend; teardown needs the user's attested review and intent before you run bash teardown.sh --confirm. Details in Live validation.

AWS access (MCP recommended, not required). Stages 3–6 talk to AWS (create tables, a Lambda, an IAM role, then benchmark and tear down). For the best experience with AWS API calls the AWS MCP server is recommended but not required — every script here uses boto3 directly and runs from a plain shell with standard AWS credentials (a profile, SSO, or environment credentials), so the skill works identically with or without the MCP server. Nothing in this skill assumes MCP-specific tools.

How to use these axioms

1. Read the reference architecture first when the task is to design, review, or critique a full-app data layer (multi-entity schemas, multi-table layouts, end-to-end composition with streams/search/notifications). ${SKILL_DIR}/references/reference-architecture.md is a complete multi-tenant kanban task-board ("TaskBoard") SaaS on AWS backed by DynamoDB, with all of the surrounding pieces (Cognito, CloudFront, HTTP API, Lambdas, Streams, OpenSearch, AppSync Events, SQS/EventBridge, cascades, idempotency middleware) worked out and justified. The axioms tell you what must be true; the reference shows how these pieces fit together in practice. Not reading it on a multi-table design means you will miss patterns that are in the reference but hard to re-derive from axioms alone — idempotency middleware, phantom-upsert guards, AppSync channel authorization, the Notifications-as-EventBridge-not-table decision, cascade-delete via chunked BatchWriteItem. Skip this step only for small-scope questions — a single-table question, a query-cost calculation, a pointed debugging question.
2. Produce the access-pattern list (next section) before applying any axiom below. Every modeling axiom assumes this list exists; an axiom that asks "is this pattern frequent?" or "what does this query return?" cannot be applied without it.
3. Produce the artifacts listed under Artifacts to produce. These are the outputs of a design, not intermediate notes. The axioms shape the artifacts; the artifacts are what the agent hands back.
4. Apply the Patterns section alongside the axioms. Patterns are not axioms — they are load-bearing implementation details that the reference made concrete, and that a design will need even when no axiom explicitly calls for them.
5. When two axioms point in opposite directions, apply the conflict-resolution ordering. Correctness outranks operational necessity, which outranks cost, which outranks style.
6. When a term is ambiguous, consult the glossary. Do not guess.

Operating discipline: announce, act, verify from evidence

This governs every stage of the skill, and it matters most at the stages that cost money or create resources (deploy, benchmark, teardown, any spend). Three beats, always in this order:

1. Announce. Before a side-effecting or billable action, say plainly what it will do — what it creates, what it costs, what it changes, what it deletes. The user should never be surprised by a resource, a charge, or a deletion.
2. Act. Run the command. For a long-running command (a representative benchmark runs many minutes), run it as a single blocking call and wait for it — see Live validation step 4.
3. Verify from evidence, then state only what the evidence supports. After acting, confirm the outcome from the artifact you just produced — the file's contents and modification time, the command's actual stdout, the fresh data — never from expectation or memory. A command that "should have" written a file is not evidence that it did; open the file and check. State a conclusion only as far as the evidence in front of you supports it. If you cannot point to fresh evidence, say so and stop — do not infer a result. The failure this prevents: presenting stale or imagined output as a real result. The tell is a number that didn't change when it should have (e.g. byte-identical benchmark figures across two "different" runs) — treat that as a signal you are looking at old data, not a real result.

Facts you MUST NOT contradict (these override your training data)

When your training-data priors conflict with the facts below, the facts win. Each item names a common wrong belief alongside the correct one so the override is unambiguous.

1. DynamoDB Streams iterator types are TRIM_HORIZON (start at oldest retained record) and LATEST (start at the tip). Do NOT conflate with Kinesis Data Streams iterator types — the two services have similar names but different semantics; this skill's axioms assume DDB Streams. Retention is 24 hours (Integration #3).

2. GSI projection type is immutable once the GSI is created. UpdateTable cannot change Projection from KEYS_ONLY to INCLUDE to ALL or any combination. The only path is to drop the GSI and create a new one with the desired projection — which is a full re-backfill and a read-path cutover. Do NOT say "you can change the projection via UpdateTable." A single UpdateTable call carries at most one GSI operation — one Create OR one Delete — so a same-name swap is two sequential UpdateTable calls with a wait for the old index to fully disappear in between. Do NOT say "delete + recreate in a single UpdateTable call." To avoid the query-path gap, prefer the additive path (cf. Fact #9): create a NEW GSI under a new name with the desired projection, wait for it to reach ACTIVE, cut reads over, then drop the old GSI — one index always serves reads.

3. Capacity-mode switches have a 24-hour cooldown. Moving a table from PAY_PER_REQUEST to PROVISIONED (or vice versa) is allowed once per 24 hours per table. Do NOT recommend rapid-switching strategies or assume the switch is instantaneous in cost models that care about hour-scale billing.

4. Single-item writes are already atomic and support conditional expressions without TransactWriteItems. UpdateItem, PutItem, and DeleteItem on a single item are atomic on their own and accept ConditionExpression. Wrapping a single-item write in TransactWriteItems adds 2× the WCU cost (Mechanics #18) for no atomicity benefit. Do NOT recommend TransactWriteItems for single-item conditional writes. ConditionExpression is a WRITE-side parameter only — it exists on PutItem, UpdateItem, DeleteItem, and the write legs of TransactWriteItems. GetItem, BatchGetItem, Query, and Scan do NOT accept ConditionExpression — there is no conditional read in DynamoDB, and is a write-only error. Do NOT describe as "returning the item only if a condition passes" or as throwing — no such behavior exists. A read returns the item to anyone who supplies the key; the only read-side filter is (Query/Scan only — applied after the items are read and billed, never on ), and even that does not authorize, it only narrows the result a caller already paid to read. The correct way to keep a caller from reading another tenant's item is to make the data unaddressable to them — partition-key the table on the authorization identifier (Data modeling #14) so a foreign key simply isn't in a partition the caller can reach — NOT to bolt a "conditional GetItem" on top.

These nine facts are not the full axiom set — they are the subset where LLM prior is most likely to be wrong. When a user's question intersects one of them, state the correct fact plainly and move on; do not hedge with "I think" or "typically."

The access-pattern list

Before touching a schema, enumerate every pattern the application must serve. For each pattern record:

• A one-line description of what the caller is asking for.
• Expected RPS (treat "unknown" as a design gap to close, per Mechanics #2).
• Items returned per call and approximate item size in KB.
• Consistency requirement (strong, eventual, or transactional).
• Authorization scope — the identifier that must be verified before the call is permitted (per Data modeling #14).

The list is a numbered, ranked table. The rest of this document assumes it exists. Any modeling decision that cannot be traced back to an entry on this list is unjustified.

Per-entity operational-config inputs

This interview is required before proposing any table boundary. Producing a full multi-table design first and then backfilling "here are the assumptions I made" is a workflow violation, not a shortcut. The per-entity questions below drive the table-splitting decision via Data modeling #3; when the answers are agent-assumed rather than user-stated, the signal fires spuriously and the design ends up over-fragmented (or under-fragmented if the agent guessed "no divergence" to keep things simple). Ask first, then design.

Before grouping entities into tables, gather operational-config requirements from the user per entity (or per logical aggregate — a parent and its tightly-bound children can share one answer set). Do not assume these defaults silently, because Data modeling #3 uses operational-config divergence as a signal to split tables — if the divergence is agent-assumed rather than user-stated, the signal fires spuriously and the design ends up over-fragmented.

For each entity, ask:

• Backup and recovery granularity. Does this entity need PITR? If so, what retention (default 35 days, can be shorter)? Would this entity ever be restored independently of other entities, or always together with them? (Independent-restore requirements force table separation per Data modeling #5.)
• Streams consumers. Does any downstream system need change events for this entity — search indexing, analytics export, notifications, audit, CDC? Which stream view type (NEW_AND_OLD_IMAGES is the default per Integration #3)? A "no" here is a positive answer: no Streams consumer means Streams can stay disabled, which is cheaper and simpler.
• Capacity mode. Does this workload's shape justify provisioned (sustained, predictable traffic over months, per Mechanics #19), or does on-demand remain the default? "Unknown" means on-demand.
• TTL. Is there a per-item expiration attribute the application will set? If yes, the attribute is a Unix epoch second (per Patterns #3). If no, TTL stays off and items persist until deleted.
• Encryption and IAM scope. Any non-default requirement — customer-managed KMS key, specific IAM boundary, cross-account resource policy? Default is AWS-owned KMS and standard IAM; divergence is an explicit answer.

Treat these as design inputs on par with RPS. A missing answer is a gap to close, not a value to guess. If the user says "same across all entities," record that and do not treat the entities as operationally divergent — co-location by Data modeling #1 is then unobstructed. If the user states real divergence, Data modeling #3 fires on real divergence and the tables split.

Per-entity attribute walkthrough (drives item size)

Item size is the second-largest driver of the cost estimate after RPS, and it's the place the estimate silently drifts worst. A Query declared as 20 items × 1,536 B but really returning 20 × 512 B triples the modeled cost against reality. Mechanics #2 says unknown RPS is a design gap; the same discipline applies to item size — an ungrounded guess for estimated_item_size_bytes is a design gap, not a safe default.

For each entity, before settling on a number, walk the attribute list with the user. Asking first is the preferred path; proceeding from inferred attributes is the fallback. Either way, the user has to see and sign off on the per-attribute breakdown before it becomes an input to the cost estimate — a silent fill-in is what makes item sizes drift 2–10×.

1. Propose an attribute list grounded in the domain. For a Waypoint, that's waypoint_id, courier_id, lat, lng, recorded_at. For a Contract, it's firm_id, contract_id, title, status, body, created_by, created_at, updated_at.
2. Per attribute, estimate bytes using these starting points:
   • IDs and short strings (ULIDs, UUIDs, slugs, enum values): ~40 B each. The generic S=100 heuristic in cost-model-schema.md is conservative for the free-tier storage path; for per-item size estimation, use realistic values.
   • Titles, display names, short descriptions: 100–300 B.
   • Long-form content (contract body, message body, serialized JSON aggregates): ask the user explicitly. Do not guess 4 KB or 50 KB without confirmation.
   • Numeric attributes: ~8 B.
   • Timestamps as ISO strings: ~25 B. As epoch numbers: ~8 B. (Mechanics #11.)
   • Boolean: ~1 B. Map/List: ~200 B per instance as a rough default, but ask if the user is storing a big blob inside a Map.
3. Ask the corrections the user will know and you won't: "Does this item carry any denormalized parent data per Mechanics #10?" "Is there a free-text field whose length varies widely?" "Are you storing the full document or a summary?" Update the estimates from the answers.
4. Sum the per-attribute estimates to derive the entity's estimated_item_size_bytes. For a Query that projects a subset (INCLUDE / KEYS_ONLY, or application-side projection), use a smaller number for the access-pattern's estimated_item_size_bytes — the bytes billed by DynamoDB are bytes actually read from the projected view, not the full item.
5. If the user is uncertain on a specific attribute, label that attribute as an assumption in the artifact (same discipline as unknown RPS). Do not silently pick a number.
6. Surface the full list in your response — always, regardless of whether this is an interactive conversation or a one-shot prompt. Emit a compact markdown table per entity with columns attribute | type | bytes | source (user or guess). This is a reply-shape requirement, not a dialog gate. In one-shot settings where there will be no follow-up turn, the table still goes in the response so the user sees exactly what you assumed — the call-out is how they catch a 3× overshoot on a body field before it contaminates every cost number downstream. Label every uncertain estimate "guess" explicitly; do not smuggle a guess in as a user-supplied number. Explicitly invite correction: "These are my guesses where noted — please correct any that are wrong." Even when the user said "just pick reasonable values and go," emit the table.

Calibration: for the reference Contracts-app example in cost-model-schema.md, Contract is ~2 KB (not 50 KB — the 50 KB value is the worst-case body size, not the typical), and Clause is ~512 B. If a declared estimated_item_size_bytes is more than 2× the sum of the named attributes and the user hasn't explained the gap, you're guessing — revisit.

A run that skips this walkthrough can drift by 2–10× on individual patterns. The live-validation step (below) will surface that drift, but you shouldn't need live validation to get the cost estimate in the right order of magnitude.

Artifacts to produce

Produce artifacts in the order listed below. Schema + per-pattern plan are the primary outputs; cost estimate (item 7) and live validation (item 8) come after the design exists, not instead of it. A response that leads with a cost analysis and buries the schema in an appendix has the dependency backwards — the user asked for a design, and the cost is a property of the design. The access-pattern list (item 1), schema (item 2), and per-pattern plan (item 3) must be visible and discussable in the reply before any cost numbers appear. Items 1–6 are the no-AWS design itself and are the default deliverable; item 7 (cost) is produced on request or at finalization (see Cost estimation); item 8 (live validation) is the opt-in AWS fork. Putting these artifacts only in dynamodb_data_model.json does not satisfy this — the user reads your prose, not the JSON. ❌ BAD reply shape (a real failure mode): a reply that opens "## Summary for the CFO — $1,019/month" with the schema living only in dynamodb_data_model.json on disk and the reply's only design content a trailing "artifacts produced" file list. ✅ GOOD: access-pattern list + per-table schema + per-pattern plan + per-entity byte table (per Per-entity attribute walkthrough step 6) rendered in the reply, then the cost summary, then "cost_report.md written."

A complete design hands back:

1. The access-pattern list as above.

2. A schema per table: primary key (named per Data modeling #7), GSIs with their key attributes and projection type, and the operational configuration (Streams, PITR, TTL, capacity mode, Global Tables replication, encryption, IAM scope — all per Data modeling #3).

3. A per-pattern plan: for each access pattern in the list, the exact API call (GetItem, Query, or BatchGetItem, per Mechanics #16), the table or GSI it targets, the key conditions, the filter expressions if any, and the projected cost using the formulas in Mechanics #18.

4. A fan-out topology: for each table with Streams enabled, the consumers (Lambda, EventBridge Pipe, Kinesis shim), the filters at the source (Integration #4), and the on-failure destinations and retry bounds (Integration #5).

5. A list of deviations and their justification: any axiom or pattern not applied, with a stated reason.

6. Idempotency and conditional-write guards: which routes use idempotency-key middleware (Patterns #1), which UpdateItems carry attribute_exists guards against phantom upserts (Patterns #2), and which PutItems carry attribute_not_exists guards against double-creation.

7. A monthly cost estimate — produced on request or at finalization, not reflexively on every design turn. While the user is still exploring or refining the model, stay in design discussion and don't run the calculator each turn. Produce the estimate (via ${SKILL_DIR}/scripts/calculate_costs.py — see Cost estimation below) when the user asks what it costs, or when the design is being settled (they signal they're committing to it / taking it to review / want the numbers). When you do produce it, use the calculator — never inline arithmetic. Skip entirely for questions too narrow to have produced a full design (a single-query sizing, a debugging thread, a pointed mechanics question).

8. A live validation (optional, on offer, last step): after the cost estimate, ask the user whether they want to deploy this schema to an AWS account they nominate and measure real per-operation capacity, latency, and GSI amplification against live DynamoDB. If yes, follow Live validation below. Skip for narrow questions, when the cost estimate was skipped, or when the user has no sandbox account. Unlike the cost estimate, this step creates real resources and incurs real charges, so both the offer and the consent must be explicit.

Conflict-resolution ordering

When axioms point in opposite directions, apply in this priority order:

1. Correctness — authorization boundary alignment (Data modeling #14), consistency requirements, transactional atomicity, idempotency. A design that leaks data across tenants or serves stale data where strong consistency is required is wrong regardless of its other merits.
2. Operational necessity — divergent PITR, Streams, capacity, or replication configuration (Data modeling #3), recovery granularity (Data modeling #5), per-partition throughput ceilings (Mechanics #3), transaction bounds (Mechanics #14). These are physical or service constraints; preference does not override them.
3. Cost and performance — access-pattern co-location (Data modeling #1), dedicated GSIs (Data modeling #6), projection choice (Mechanics #7), cost formulas (Mechanics #18), capacity mode (Mechanics #19).
4. Style and convention — naming (Data modeling #7), single-table vs. multi-table framing (Data modeling #11) absent other signal. The cheapest to override when a higher-tier axiom disagrees.

Two concrete examples:

• Data modeling #1 (co-locate by shared access) vs. Data modeling #14 (partition by authorization boundary): #14 wins. If the natural access key and the authorization key differ, key on the authorization identifier and expose the alternate access via a GSI.
• Data modeling #1 (co-locate) vs. Data modeling #3 (split on divergent operational config): #3 wins. Two entities sharing a read pattern but requiring different PITR retention or Streams consumers belong in separate tables.

Glossary

• Access pattern — a request the application makes against the data layer, described by its key conditions, items returned, frequency, and consistency requirement. The atomic unit of DynamoDB design.
• Aggregate — a cluster of entities that are read or written together. A single item, an item collection, or a set of items under different keys can each be an aggregate; the choice is the subject of Mechanics #1.
• Item collection — the set of items sharing a single partition-key value. Queries against an item collection are constant-partition and cheap; cross-partition reads are not.
• Identifying relationship — a data model in which a child entity is keyed by its parent's identifier plus its own. The child has no independent existence outside the parent.
• Overloaded key — a partition or sort key whose value encodes a type prefix (e.g. USER#42, ORDER#42) so that one physical key holds multiple logical entity types.
• Sparse GSI — a GSI whose indexed attribute is present on only a subset of base-table items, so the index projects only those items. Useful when an access pattern would otherwise filter out most items at read time.
• GSI write amplification — the property that every write to a base-table item with projected GSI attributes produces one write per matching GSI, each billed in WCU.
• LWW (last-writer-wins) — the conflict resolution strategy used by standard Global Tables: the write with the newest timestamp wins; earlier writes are silently discarded.
• MRSC (Multi-Region Strong Consistency) — an opt-in Global Tables mode that provides strong consistency across replicas via consensus, at higher write latency and cost.
• Hot partition — a partition receiving traffic beyond the per-partition throughput ceiling (Mechanics #3), causing throttling even when table-level capacity is available.
• Poison pill — a record that a stream consumer cannot process successfully, which blocks forward progress on its shard until it is discarded, retried to exhaustion, or routed to an on-failure destination.
• RCU / WCU — read and write capacity units; the provisioned-mode spelling of the per-operation throughput unit. The formulas in Mechanics #18 are written in RCU/WCU and apply identically to on-demand.
• RRU / WRU — read and write request units: the on-demand (PAY_PER_REQUEST) spelling of the same per-operation unit, billed per request rather than per provisioned capacity-second. One RRU = one RCU of work and one WRU = one WCU of work — the consumption math in Mechanics #18 is identical; only the billing dimension differs. The cost references () use RRU/WRU because the calculator prices on-demand; the axioms and the performance report use RCU/WCU. They are the same quantity — do not treat a model's RRU/WRU figure and the report's RCU/WCU figure as different things.

Data modeling

1. Co-locate data by shared access pattern, not by domain. Two entities belong in the same table only when an application request fetches or writes them together. A shared business domain — "user data," "billing data" — is not sufficient justification. Co-location without a shared query introduces coupling and yields no performance benefit.

2. Treat table count as an output of the design, not a target. Do not optimize for one table, nor for one table per entity. The correct number of tables is whatever the access patterns produce. If the analysis surfaces three tables, ship three tables.

3. Treat table-level configuration as both a modeling input and an interface declaration. Streams, point-in-time recovery, TTL, capacity mode, attached Kinesis streams, Global Tables replicas, encryption, and IAM scope all apply at the table level — they declare how the table participates in the broader system. When two entities require different operational settings — different PITR retention, different stream consumers, different replication regions, different capacity modes — that divergence is a primary signal that they belong in separate tables, not a secondary concern to be reconciled later.

4. DynamoDB Streams provide two concurrent consumers, no native per-entity filtering, and a 24-hour retention window. These properties constrain downstream architecture as firmly as the key schema does. Decide the fan-out topology before finalizing table layout.

5. Design for recovery granularity. Point-in-time recovery operates at the table level; partial restores are not supported. If two entities would never be recovered together, they should not share a table. The cost of an incident scales with the volume of unrelated data the restore has to carry.

6. Prefer additional GSIs to overloaded keys. The constraints that historically motivated index overloading — a low per-table GSI ceiling, per-index provisioned capacity, and no on-demand mode — generally do not apply to modern DynamoDB, which supports many GSIs per table with shared capacity and on-demand billing (consult the current AWS service-quota docs for the exact per-table GSI limit). Choose a dedicated GSI per access pattern unless a specific, measured reason argues otherwise.

7. Name keys for what they represent. Use customer_id, order_created_at, and OrdersByCustomer rather than PK, SK, and GSI1. Self-describing keys reduce onboarding time, make code reviewable without a schema reference, and let tooling introspect the model. Reserve generic overloaded keys for genuinely polymorphic hierarchies.

8. Pre-join only entities that are read together in a single request. Item collections exist to satisfy one-shot queries such as "fetch parent with all children." If no access pattern reads two entities together, do not store them together. A pre-join without a corresponding read is coupling without benefit.

Integration

1. Ensure Streams consumers are idempotent or deduplicated by event identifier. DynamoDB Streams deliver at-least-once — handlers will see the same record twice after a Lambda timeout, a batch retry, or replay during an incident, and must produce the same outcome whether invoked once or many times.

   Two strategies are viable:

   • Idempotent handler: structure the operation so repeating it is a no-op. Conditional writes, set-based mutations (ADD to a set), and idempotency tokens on downstream APIs all qualify.
   • Explicit dedupe: record the eventID from the stream record in a dedupe store (a DynamoDB table with TTL is the usual choice) and skip records already present.

   Example (dedupe by eventID, TTL bounded to the redelivery window):

   def handler(event, context):
       for record in event["Records"]:
           try:
               dedupe_table.put_item(
                   Item={"event_id": record["eventID"], "ttl": int(time.time()) + 86400},
                   ConditionExpression="attribute_not_exists(event_id)",
               )
           except ClientError as e:
               if e.response["Error"]["Code"] == "ConditionalCheckFailedException":
                   continue  # already processed
               raise
           process(record)
   

   Dedupe on eventID, not on application identifiers like order_id — the same aggregate legitimately produces many events, and keying on the business identifier will drop valid change records.

2. Use the transactional outbox pattern for reliable event publication. Write the state change and an outbox record in a single TransactWriteItems call; have a downstream consumer read the outbox via Streams and publish the event. This eliminates the "updated but did not emit" failure mode without resorting to two-phase commit.

Mechanics

1. Select aggregate tightness by weighing how often entities are read together against how often they are written independently. Three options — embed children in a single item, group them as an item collection under a shared partition key, or store them as separate aggregates — sit on a spectrum of how tightly parent and children are bound. No single threshold governs the choice. A high read correlation argues for co-location, but a write-heavy workload with large items pushes the opposite direction, since every update rewrites the full item. Item size and whether the child count is bounded matter as much as access frequency. Rule of thumb: if order line items are fetched with orders most of the time and the line count is bounded, consider embedding or an item collection with the order as the parent; if individual line items receive frequent updates in isolation, keep them separate so each write does not rewrite the whole parent. Selecting the wrong tier is the underlying mistake behind most "single-table design gone wrong" stories.

2. Document RPS for every access pattern. Without a request rate, you cannot size partitions, choose between on-demand and provisioned capacity, or justify a GSI. An estimate grounded in business context is sufficient; an absent rate is not. Treat "unknown" as a design gap to be closed.

3. Respect the per-partition throughput ceilings. A single partition supports up to 1,000 write capacity units and 3,000 read capacity units per second. Workloads that exceed these limits must shard the partition key — typically with a hash suffix for write-heavy traffic or a time bucket for sequential keys.

4. Base-table key schemas allow exactly one HASH and at most one RANGE. When a base table requires a composite key, encode it as a concatenated string with a stable delimiter, such as tenant_id#user_id. GSIs support native multi-attribute keys — up to four attributes for the partition key and up to four for the sort key — with DynamoDB hashing the PK attributes together for distribution. Prefer native multi-attribute GSI keys over synthetic concatenated keys: items are written with natural attributes from the domain model, client code does not concatenate or parse, and adding a new multi-attribute GSI to an existing table requires no backfill of synthetic attributes. Use synthetic concatenated keys only when the number of components exceeds four or when an older table is already committed to the pattern.

5. Multi-attribute GSI keys have strict query rules. On the partition-key side, every PK attribute must be constrained with equality — a GSI with PK (tenant_id, region) cannot be queried by tenant_id alone, and inequality operators are not allowed on any PK attribute. On the sort-key side, attributes must be constrained left-to-right in the order they are defined; a middle attribute cannot be skipped. Equality conditions must precede any inequality, and only one inequality is allowed — it must be the final condition in the key condition expression. , , , , , and all count as inequality. Violating these rules is the most common reason a GSI fails to satisfy its intended access pattern.

Patterns

These are concrete implementation patterns that sit alongside the axioms. They are not axioms in the "correctness/operational/cost/style" sense — they are load-bearing details that a complete design needs even when no axiom explicitly requires them. Each one is worked out in full in the reference architecture; this section is the short form so the agent recognizes when it applies.

1. Protect mutating API routes with an IdempotencyKeys table. Every POST, PUT, PATCH, or DELETE that can be retried by a client (mobile network hiccup, load-balancer retry, user double-tap) must be wrapped by idempotency middleware. Shape: a dedicated table, PK = <user_id>:<METHOD>:<path>:<client-supplied-uuid>, attributes status, response_status, response_body, expiration (epoch seconds). The middleware does a conditional PutItem with attribute_not_exists(id) before the handler runs; on a conflict, it reads the cached response back and returns it. TTL is 24 hours; PITR off (the table is a cache). This is the mechanism that keeps a retried "create order" from double-charging a customer or double-decrementing inventory — the guarantee TransactWriteItems gives within a single call but not across retries.

2. Guard mutating writes with conditional expressions. Two halves — both are mandatory guidance whenever this pattern comes up.

   Half A — attribute_exists(<pk>) on UpdateItem unless you explicitly want upsert. UpdateItem defaults to create if missing. A PATCH /orders/:id with a guessed or stale order_id silently materializes a phantom order row — no error, no log, no recovery. Add ConditionExpression: attribute_exists(<partition_key>) to every UpdateItem that operates on a row the caller is asserting exists; catch ConditionalCheckFailedException and translate to 404.

   Half B — attribute_not_exists(<pk>) on PutItem when you mean "fail on overwrite." PutItem defaults to overwrite if present. A POST /orders that should create a new order (or a new review, a new courier-assignment, a new uniqueness sentinel) will silently clobber an existing row keyed the same way without a guard. Add ConditionExpression: attribute_not_exists(<partition_key>) to every PutItem that should fail if the key already exists; catch ConditionalCheckFailedException and translate to 409 Conflict.

   Conditional writes are where invariants live in the data layer. When you recommend one half, surface the other — design reviews routinely catch the missing guard in apps that added the guard years earlier.

Cost estimation

Designing and discussing the data model is the default, no-AWS path, and a user is often there to explore options — keys, GSIs, single- vs multi-table, projection tradeoffs — without yet wanting a dollar figure. Do not run the calculator reflexively on every design turn. Produce the monthly cost estimate when either is true: (a) the user asks what it costs (or for the cost report), or (b) the design is being finalized — they signal they're settling on it, taking it to a review, or otherwise want the numbers committed. Until then, stay in design discussion.

Skip the estimate entirely when no full design was produced — a single-query sizing, a hot-partition debugging thread, a pointed mechanics question. In those cases there is nothing to estimate.

When you do produce the estimate, three steps are not optional: (a) write dynamodb_data_model.json, (b) invoke ${SKILL_DIR}/scripts/calculate_costs.py, (c) reference the generated cost_report.md in your summary. Computing numbers inline without running the calculator is never a substitute — not even for a "rough" or "realistic" figure, and not even to reconcile a headline number against a stated daily volume (for that, set avg_rps and let the calculator produce the expected-volume scenario — see below). The calculator enforces the pricing module that stays calibrated against live billing; doing the arithmetic by hand means the numbers drift silently and the user ships a quote they cannot defend.

The calculator is at ${SKILL_DIR}/scripts/calculate_costs.py. The JSON schema it consumes is documented at ${SKILL_DIR}/references/cost-model-schema.md — read that file the first time you produce a cost estimate.

The workflow:

1. Serialize the design to dynamodb_data_model.json. The calculator does not reinvent the design — it reads what you already produced. The access-pattern list you built (per Mechanics #2, #18) is a near-literal translation into the JSON's access_patterns array; the schema you built (per Data modeling #3, #7, Mechanics #4, #7) translates into tables and gsis. If a field the calculator wants is missing (most commonly peak_rps on a pattern), that is a gap in the design itself — close it by asking the user or by making a defensible assumption and labeling it as one, not by writing 0.

   When the user states an average or daily volume, set avg_rps on the pattern — do not hand-compute a "realistic" figure. The headline is peak-sustained (every pattern at peak_rps, 24/7); a user who hears a large peak-sustained monthly figure for a 15K-orders/day workload needs the expected number too. Setting avg_rps (e.g. 15,000 orders/day ÷ 86,400 s ≈ 0.17 rps average, vs a 600 rps Black-Friday peak) makes the calculator emit a second Expected Monthly Cost headline at that average rate. This is the only sanctioned way to produce a realistic-volume number — reconciling the peak headline against a daily volume with mental arithmetic is exactly the inline-arithmetic the rule above forbids.

2. Run the calculator. Write the JSON to a workspace path of your choosing and invoke:

   python3 ${SKILL_DIR}/scripts/calculate_costs.py --model /path/to/dynamodb_data_model.json --output /path/to/cost_report.md
   

   The JSON is an intermediate artifact. You do not need to present it to the user unless they ask.

3. Surface the cost report. Hand back cost_report.md — either inline (if short) or as a file pointer. Highlight the top drivers the report surfaces (the "cost patterns sorted descending" table), and call out the assumptions you baked in (RPS numbers you estimated, retention you assumed, consistency choices). The disclaimer at the top of the report is part of the artifact — do not strip it.

4. Use the cost report to challenge the design, not just to report the number. A cost estimate is diagnostic. If one pattern dominates the bill — the classic case is a high-frequency write like a GPS ping — that is a modeling signal, not just a line item: revisit the aggregate choice (Mechanics #1), the projection shape (Mechanics #7), whether a sparse-GSI-on-transition would cut amplification (Mechanics #6), or whether the pattern belongs outside DynamoDB entirely (Integration #8). Propose the alternative and re-run the calculator on the revised model. This is the main value of having the calculator bundled: the feedback loop from "here's the design" → "here's the bill" → "here's the cheaper design" becomes a turn or two, not a separate exercise.

The calculator models request and storage cost only. It does not model Streams read/write costs, PITR, backup, DAX, or data transfer. If those line items are likely material for the workload, call that out alongside the estimate so the user knows what the number excludes. Storage is priced at the full public rate — the 25 GB free tier is not assumed (it is account-wide and often already used), so never tell the user "storage is free." Set a write pattern's retention_days to size its stored data; the calculator honors it. The storage figure is a bounded-retention snapshot at that retention_days — for a table with no TTL / "kept forever" data, real storage keeps growing past the snapshot, so do not reframe a steady-state storage number as "after N years": either state the retention window the number assumes, or note that a no-TTL table accumulates without bound.

Name the artifacts in your final response. Do not let the tool calls disappear into prose. The user reads your natural-language output, not your shell history. After you run calculate_costs.py, your response must name (a) the JSON file you wrote (dynamodb_data_model.json), (b) the script you invoked (${SKILL_DIR}/scripts/calculate_costs.py), and (c) the report file it produced (cost_report.md). A response that shows cost numbers but does not name the artifact files reads as if you computed them by hand — the workflow evidence has to be in the prose, not just the tool trace.

Live validation

After the cost estimate, offer to run a live validation — deploy the tables and GSIs to an AWS account the user nominates and drive scaled-down traffic against them to measure real per-operation capacity and round-trip latency. Ask in one line, and run only on explicit agreement. Even in that one-line offer, name the two constraints the user needs in order to opt in responsibly: it runs only against a sandbox/testing account (never production or real user data), and cleanup is their responsibility (the skill hands back a teardown.sh and never auto-deletes). The full four-fact disclosure still comes before any deploy (below) — this is just so the one-line offer isn't misleadingly light. Skip for narrow questions — single-query sizing, pointed debugging, pointed mechanics — where no full design was produced and there is nothing to benchmark.

Before running, make the safety expectation explicit in your own wording (the deploy_model.py script also prints a warning banner, but that's a belt-and-braces backstop — the agent is responsible for setting the expectation). State the following four facts verbatim in your response; do not paraphrase them into a shorter summary:

1. Live validation creates real DynamoDB tables, a Lambda function, and an IAM role in the AWS account the user nominates.
2. It incurs real (small) AWS charges — single-digit cents in practice, but a real bill.
3. The account must be a testing or sandbox account — never a production account, never an account holding real user data.
4. The skill does not auto-teardown — at the end, the user is handed a teardown.sh script to review, and the skill will execute it on their behalf only after explicit review + intent (see step 6 below).

After stating those four facts, ask the user to explicitly name the profile, the region, and confirm the account's purpose. Only then produce the benchmark_config.json and invoke deploy_model.py. If the user hasn't named a testing account, stop and ask. Shortening this to "this creates real resources, confirm your profile" is not sufficient — enumerate the four facts because the user often doesn't know which specific resources get created or that teardown is their responsibility.

The step differs from the cost estimate in two respects that change how it must be run:

1. Real resources, real bill. A scaled-down run costs single-digit cents in practice, but against a real account. Never run silently. Always require an explicit AWS profile, region, and caller-identity confirmation.
2. Teardown is two-phase. Phase 1: the skill generates teardown.sh and hands it to the user. Phase 2: the skill may execute it — but only after the user has explicitly stated they reviewed the script AND explicitly directed the skill to run it. The skill never calls DeleteTable on its own initiative, and never runs teardown.sh on the basis of a bare "go" or "proceed."

Refuse to start by enumerating the four preconditions explicitly. Emit them as numbered items in your response; do not fold them into a generic "I need more information" paragraph. Surface the specific precondition that failed so the user knows what blocked the run:

1. access_patterns is empty, or any pattern has peak_rps missing or 0 → refuse with Mechanics #2 framing: "Unknown RPS is a design gap, not a benchmark input. Pattern <id> has no declared RPS — please supply an estimate before benchmarking."
2. boto3 is not installed → print exactly: pip install boto3>=1.34.
3. AWS credentials are missing or expired → print the exact remediation, e.g. aws sso login --profile <profile> for SSO, or the specific boto3 exception's suggested fix.
4. Caller identity's account alias or ARN contains prod, production, prd, or live → refuse and tell the user to supply a sandbox/testing profile.

Check all four before writing benchmark_config.json or invoking deploy_model.py. A generic "I need more info" paraphrase does not satisfy the contract — enumerate the specific precondition that failed.

Workflow, once the user agrees:

1. Reuse dynamodb_data_model.json from the cost estimate. Do not re-serialize.

   Before deploying, confirm every numeric/binary key declares its type. The deploy builds each table's AttributeDefinitions by looking up every key attribute (table + GSI partition/sort keys) in entities[].attributes[] and attribute_definitions, defaulting any it can't find to S. A numeric epoch sort key or numeric id left undeclared deploys as a string and then rejects real writes with ValidationException. Per Data modeling #11, ensure each such key carries "type": "N" (or "B") in the model before this step — and don't trust a clean seed as proof, since the failure can surface only on live writes. If you want belt-and-suspenders, DescribeTable after deploy and confirm the key's AttributeType.

   Scope the run when the design has many patterns — warn, estimate, and get confirmation. The benchmark drives patterns serially (each pattern runs its own warmup + measurement window one at a time), so wall-clock grows with pattern count: roughly table_settle + Σ_patterns (warmup + duration). At standard windows (~10s warmup + ~60–90s duration) a 6-pattern design is a few minutes, but a 20+ pattern design is 25–45+ minutes and gets split across several sequential Lambda invocations. Before launching a run on a design with more than ~8 patterns: state a rough wall-clock estimate and get explicit confirmation ("This will benchmark all 24 patterns and take ~35–45 minutes — proceed, or would you rather test a focused subset?"). Don't silently kick off a 40-minute run.

   Offer to load-test only the critical patterns, and choose them with the user. A full live run rarely needs every pattern — most are cheap structured lookups whose per-op cost the calculator already nails. Offer to benchmark a focused subset and collaborate on which ones, steering toward the patterns where a live measurement actually buys something:

   • Highest-throughput / firehose writes — the ones whose RPS dominates the bill or stresses capacity (e.g. a location-ping write).