spencer
Community@shuddl · los angeles, ca
dramatic!
Agent Skills by spencer
Showing 16 vetted skills indexed across 1 GitHub repositories.
complete-append-only-insert-guards
Validates D1 BEFORE INSERT guard triggers against every unique conflict key on append-only tables.
keep-readmodel-consistent-with-ledger
Ensures every read-model projection stays consistent with the append-only event ledger.
park-unroutable-work-never-destroy-it
Guides queue consumers to retry unroutable messages toward a DLQ instead of acking them.
fail-closed-on-inherited-visibility
Enforces fail-closed handling when ledger events inherit visibility from referenced events.
share-lint-matchers-with-parity-tests
Extract shared regex fragments across duplicate lint scanners and prove parity with one probe corpus.
make-agent-idempotent-and-adapter-ported
Implements deterministic id generation and three-adapter ports for at-least-once Cloudflare Queue consumers.
gate-external-routing-server-side
Wraps external routing APIs behind an OSRM-shaped server-side port for advisory travel estimates.
enforce-server-side-gate-parity
Enforces server-side authorization gates on API routes that bypass the sequencer when persisting physical facts.
keep-map-instrument-truthful
Fixes exception pulse parity and heading rendering bugs in MapLibre fleet map layers.
reconcile-gate-sentinels-with-exit-codes
Reconciles structured gate sentinels with process exit codes using pessimistic-wins semantics.
keep-ops-record-reconciled-with-deploys
Reconciles ops documentation claims with actual deployment and provisioning state after world-changing actions.
prove-tenant-isolation-read-paths
Adds cross-tenant isolation test cases for every API route and R2/D1 key builder.
terminal-gallery-map-ui
Builds a 1,000-shipment live map canvas with Mapbox or MapLibre GL layers and a five-token greige design system.
preserve-canonical-hash-byte-law
Preserves the frozen canonical-JSON byte law governing event hashing, signing, and chain verification.
redact-counterparty-payloads-completely
Strips internal fields from counterparty-visible event payloads and API responses.
harden-agent-against-model-trust
Harden LLM-consuming agents against prompt injection with deterministic fail-closed verification rules.
Frequently Asked Questions About spencer
FAQPage SchemaWhat tasks can I accomplish with SHUDDL's skills?▼
You can enforce append-only D1 ledger guards, keep read-model projections consistent after corrections and voids, park unroutable queue messages instead of destroying them, redact counterparty-visible payloads, preserve canonical hash byte law, and build truthful live-shipment map surfaces with exception states.
Who are these skills designed for?▼
Engineers building and reviewing the SHUDDL logistics platform: backend developers writing D1 triggers and projections, queue consumer authors handling at-least-once delivery, agent builders integrating email, LLM, routing, and Stripe seams, and frontend engineers maintaining the Mapbox GL JS operational map.
How do the skills fit into a practical development workflow?▼
Each skill triggers during code authoring or review: when adding API routes persisting GPS or custody facts, editing canonical.ts or redact.ts, adding gate harnesses parsing ##SHUDDL-GATE## sentinels, or reconciling ops markdown after deploys, secret bindings, and DNS changes.
What prerequisites and dependencies do these skills assume?▼
A Cloudflare stack: D1 databases with BEFORE INSERT triggers, R2 tenant-scoped storage, Queues with at-least-once consumers, and Workers API routes. Map work assumes Mapbox GL JS or MapLibre GL JS; routing gates assume Mapbox Matrix/Directions or self-hosted OSRM/Valhalla.
How do the skills handle multi-tenant security and data visibility?▼
They mandate fail-closed inheritance of visibility attributes, complete redaction of internal fields like gl_map, division, and cost amounts from counterparty payloads, tenant-isolation test coverage for every new read path, and server-side gating of client-supplied shipment_id and device_id writes.