Work

The repository outcome is a comprehensive documentation refresh for Oliphaunt, reorganizing learning paths and reference pages around storage, runtime behavior, transactions, backups, and releases. It adds a docs authoring workflow with source maps and an audit, and introduces build-time checks that validate SDK quickstart snippets and browser/site links. It also updates the site UI layout and styles for consistent navigation and search.

2 commits · 88 files

The project strengthens trusted embedded PostgreSQL execution by enforcing a single-backend model, preserving host process boundaries, and adding strict session and interruption semantics. It also introduces bounded output and protocol transport contracts, updates host-side and guest runtime behavior accordingly, and expands smoke and contract tests to verify the new guarantees.

9 commits · 240 files

The project’s CI and local tooling now execute PostgreSQL’s official regression suite against pinned PostgreSQL 18.4 sources, including a WASIX-appropriate embedded schedule and a WASIX postmaster qualification lane. Upstream regression runs are integrated into the WASIX regression flow with one-connection execution for embedded runtimes, required input validation, and artifact uploads of results and diagnostics. A dedicated WASIX Rust test runs the upstream regression via the shared regression-

1 commit · 13 files

The site now serves technical posts under /writing instead of /blog. All canonical URLs, feed entries, sitemap entries, JSON-LD structured data, and UI links are updated accordingly. Requests to old /blog paths are permanently redirected to the new /writing routes, and work-log is similarly redirected to /work.

1 commit · 24 files

The repository adds an MDX blog post draft describing a shift to cent-pegged AI credit billing using PostgreSQL triggers and serializable transactions with Temporal for work orchestration. It also adds updated illustrative images with edited captions and alt text removing confusing numeric amounts, plus an Open Graph image. A draft-specific metadata export supports consistent social previews in the site pipeline.

1 commit · 7 files

The repository introduces a dedicated blog-images workflow that selects illustration subjects, defines an editorial cut-paper house style, and generates matching inline and social images with safe insertion guidance. It also upgrades multiple existing blog posts by adding figures and updating visuals to remove misleading labels or misleading props, including replacing metaphors with grounded scenes and adding diagrams where needed.

1 commit · 47 files

The project now includes an expanded face-motion asset bundle for the v13 resume experience, with a full manifest plus supporting atlas and transition images. It adds tooling and QA scripts to build and verify the face-motion atlas and transitions, and updates the UI and asset-loading logic accordingly. The result is a more complete, ready-to-render portrait motion set for the resume page.

14 commits · 461 files

The site’s build was updated to newer versions of Next.js, MDX tooling, Mermaid, and code highlighting libraries. Mermaid diagram type mappings now include an additional Class Diagram v2 option and an XY chart option. The Mermaid layout integration was adjusted to use the Elk layout package without the previous compatibility pin rationale.

1 commit · 3 files

The blog post page UI was simplified by removing the author link and the optional updated time display. This change reduces the amount of metadata shown above the post content and eliminates the navigation link for the author.

1 commit · 1 file

A lead finding workflow was redesigned to share evidence across research passes, delay contact research until after ranked promotion, and move web search out of model calls via an explicit retrieval step. As a result, one benchmark domain’s discovery cost dropped from about $3.89 to $0.58, using fewer web searches and less elapsed time while raising mean opportunity scores.

2 commits · 1 file

The repository refines its blog-writing prompts and expands source/documentation coverage while adjusting site-SEO guidance to focus on search discoverability, metadata, structured data, indexing controls, social previews, and agent readability. It also switches to the repository’s existing Bun-based test, lint, and typecheck validation flow and emphasizes using published-surface inspection plus Search Console/Bing data for measured outcomes rather than claiming ranking improvements from builds.

1 commit · 10 files

The site outcome is a redesigned blog article experience with consistent typography, spacing, and media layouts. Code blocks now use a unified toolbar with tooltip-backed language icons and fixed-size copy controls, while Mermaid diagrams render in a shared, token-aligned style. GitHub links generate richer repository and pull request previews using embedded Octicons and escaped content, supported by added tests for code blocks, images, and GitHub embed rendering.

1 commit · 18 files

The repository adds structured, reusable guidance for writing in a specific voice and for performing website and search readiness audits. It also introduces publishing infrastructure that keeps Markdown image references resolvable, generates full-size social title cards, and enforces preview noindex behavior. Blog pages now expose describedby metadata for llms.txt, share-image behavior is simplified, and a deployment-focused SEO audit script plus tests validate the workflow.

5 commits · 41 files

The system now reuses previously accepted summaries across compatible work context changes without triggering new paid work, and it debounces repeated requests. It records daily started-request usage with finer control, preserving remaining capacity for new work while pacing historical jobs. Database schema changes capture per-attempt identity, repository, and request start timestamps across retries and UTC day boundaries, even through projection deletion, to keep auditability and reconciliation

4 commits · 16 files

The system can recover missing or truncated GitHub text patches by fetching unified diffs, validating them against a file ledger, and rejecting stale pull-request snapshots. Summary providers now track structured error codes through database schema updates and worker/store logic, enabling bounded retries while preserving rejection reasons. Diff evidence handling was tightened for binary and metadata cases, with tests covering recovery, stale PR rejection, and retry behavior.

1 commit · 20 files

The system now generates a domain traffic surge email that records exact DNS lookup counts per owned domain and links each domain to its Marketplace and Analytics pages plus an Experiments opt-in path. Domain management deep links are preserved during hydration, auth, ownership, and remote-flag resolution, and only reset to a default tab when a requested surface is confirmed unavailable.

1 commit · 8 files

AI credits charging and refunding was rebuilt around a PostgreSQL ledger with a trigger-maintained read-only balance projection. Billing mutations now run in serializable transactions with retry on serialization failures, and ledger writes are validated by database constraints and foreign-key source references. Leadgen workflow start and outbound/studio credit endpoints were updated to use ledger-entry IDs for consistent refund behavior.

2 commits · 42 files

The repository delivers a tiny, dependency-light, type-safe codec that encodes and decodes Facebook Thrift Compact Protocol raw struct payloads using Uint8Array. It enforces stable field IDs, optionality rules, schema validation via Standard Schema, and throws a dedicated ThriftError for malformed or invalid inputs. CI now verifies formatting, linting, typechecking, regression vectors, and a Node 22/24 published surface check, with releases automated via semantic-release and npm trusted/OIDC.

2 commits · 28 files

The release tooling now treats ten-minute crates.io refill behavior correctly when no explicit caller deadline is provided, avoiding premature deferral on long Retry-After waits. Tests were updated to expect the full wait duration and verify that only the exact frozen crate bytes are replayed after HTTP 429. The publication controller now includes the frozen Cargo publishing script in its managed set.

1 commit · 3 files

The project updates its release process to grant id-token write permissions for both publish and bootstrap jobs so npm provenance can be generated. It also extends CI tests to assert that required OIDC settings are present for those jobs. Documentation is clarified on how bootstrap and subsequent publish runs should handle ledgers and checkpoint scope retries without cross-run checkpoint migration.

1 commit · 3 files

The release GitHub workflow was updated to set up Moon in the correct order relative to the CI publishing gate steps, ensuring Bun-based checks run only after the Moon installation step. A new test was added to parse the release workflow and verify that the publishing and bootstrap gate scripts occur after the Moon setup action in each relevant job.

1 commit · 2 files

The release workflow now pins and verifies publishing code using a validated publication controller, allowing reuse only when only publication-only changes are present. Publishing and bootstrap steps require successful CI and improved release-please readiness checks, while bootstrap registry identity logic now checkpoints mixed inventories and resumes safely. Additional guards ensure attestation verification tracks the actual publisher commit SHA.

1 commit · 15 files

The repository’s release and policy tests were strengthened to ensure cross-workflow artifact downloads still reference the correct CI producer job names. It also added a check that publish and publish-bootstrap install the same approved publication candidate and reject approval drift. This makes the release bootstrapping and workflow gating behavior more deterministic and regression-resistant.

1 commit · 4 files

Native and WASIX extension release tasks no longer depend on the shared extensions source fetch scope. A CI test was added to confirm the native build only materializes the native runtime source scope and not the extensions scope, while the WASIX build only materializes the WASIX runtime source scope and not the extensions scope. This tightens CI validation of which source inputs are pulled for each artifact builder.

1 commit · 4 files

Release tooling now extracts npm manifests from the exact tarball member path and tightens canonical iOS native dependency validation based on whether the artifact includes an iOS xcframework. Tests were expanded to ensure manifest discovery works when a dependency is bundled. This improves correctness and detects mismatched extension metadata during publication freezing.

1 commit · 3 files

The repository refined its quality guardrails to emphasize correctness for monetary units, billing and asset ledger transitions, and identity and authorization boundaries. It also expanded guidance for schema-focused changes and validation harness coverage, including failure paths, retries, and retryable concurrency. The result is more consistent reviews that reduce risk of incorrect charges, authorization mistakes, and state inconsistencies.

The app now uses reusable WebMCP tool bindings that derive and validate tool inputs from existing contracts, standardizing read-only and non-destructive annotations. DNS chat tools were rebuilt to load authoritative DNS state, validate proposals, and stage only accepted changes for review with structured invalid_input reporting. Search/cart WebMCP tools were consolidated into a page-local tool factory with readiness checks, abort handling, concurrency locks, and comprehensive unit tests.

1 commit · 16 files

The leadgen UI was updated so outbound outreach is always prepared via one “Prepare” action, removing the separate “Review outreach” flow and “Outreach” column. The prospect table now renders outreach content through a provided renderOutreach slot and ensures tests expect the prepare button while avoiding the old review text. UI spacing and bookmark/hidden icons and button styling were adjusted to match the new behavior.

1 commit · 3 files

The frontend updates lead reporting by adding a new responsive LeadgenProspectTable that renders as a table on desktop and stacked draggable cards on mobile. It includes reordered drag behavior, wrapped cell styling, and unique contact-email handling for outreach review drafts. The change also updates TanStack React Table usage across admin, frontend, and UI packages to the react-table-v8 API and bumps frontend to react-table v9.

1 commit · 58 files

A single deterministic backend builds validated DNS plans by combining record validation, setting validation, and conflict repair. The AI validation API now validates only after domain ownership checks, and merges record and setting refusals into one review list. The frontend coordinates the DNS chat assistant and WebMCP workspace with abortable, page-scoped lifetimes to prevent late updates, while adding sandboxed tools for staging DNS plans and domain cart actions.

1 commit · 33 files

Blog article rendering now uses a dedicated prose wrapper and updated heading typography. Portfolio and codex list rows were redesigned for consistent grid-based layouts and clearer spacing. Global styling was streamlined and emphasis/underline behavior was adjusted, while embedded GitHub cards, code blocks, and Mermaid placeholders received refined styling and animation controls.

2 commits · 29 files

Date formatting now reuses cached Intl.DateTimeFormat instances keyed by format and timezone, improving performance for repeated calls. The repository avatar rendering no longer passes an unoptimized image flag.

1 commit · 2 files

The site now fetches initial GitHub activity data server-side for both the homepage and work log, and removes suspense-based placeholders for those sections. GitHub activity days are grouped and labeled using a fixed Asia/Kolkata (IST) timezone instead of the viewer’s local timezone. A test was added to verify the generated homepage HTML includes the correct IST work and totals across UTC midnight.

1 commit · 7 files

Journey expand/collapse behavior was updated to animate height and progressively reveal details without snapping. Reveal transitions now adapt to reduced-motion settings and use consistent entrance and exit behavior even when toggled rapidly. A new automated Playwright script validates motion on desktop/mobile, keyboard and reduced-motion accessibility, and ensures content is fully revealed or hidden reliably.

1 commit · 6 files

The site’s blog and writing views were updated to use a viewer-timezone aware date-time component for more accurate timestamps. Navigation and layout were refined with new shell and header structure, including a dedicated journey page and updated route handling. A new test setup and additional date, copy-email, and structured UI coverage were added to support these changes.

2 commits · 159 files

The release workflow now classifies bootstrap needs by exactly absent Cargo and npm names, then provisions only the minimal registry tokens for that scoped absent-name ledger. It seals bootstrap using a checkpoint that preserves the approved exact publication lock and resumes correctly on reruns. Capacity gating and reconciliation now exclude existing names that await later versions, preventing forbidden bootstrap writes and ensuring deterministic release mutation behavior.

1 commit · 13 files

Oliphaunt packages and SDKs were bumped to version 0.2.0, including native runtimes, WASIX runtimes, broker helpers, and language bindings across JavaScript, Rust, Swift, Kotlin, and React Native. The update standardizes embedded PostgreSQL public APIs and aligns native and WASIX behavior. It also introduces breaking WASIX changes around storage variants and browser IndexedDB layout handling.

1 commit · 108 files

Oliphaunt versioning was bumped across runtimes, SDKs, and extensions, including a new v0.2.0 for the Rust WASIX crate and other packages. The update consolidates embedded PostgreSQL APIs, switches wasix-ts host runtimes through Rust Node-API, and removes temporary or application-data storage variants. Browser IndexedDB uses a new per-database v3 layout without migrating earlier data.

1 commit · 84 files

The repository updates the project to Oliphaunt version 0.2.0 across native, WASIX, Node-API, and JavaScript/TypeScript, Rust, Kotlin, Swift, and React Native SDKs, plus bundled extensions. It applies breaking changes around the embedded PostgreSQL API unification and the Rust WASIX storage model layout used by browser IndexedDB. Example apps and dependency manifests are aligned to the new versions.

1 commit · 108 files

The repository updates Oliphaunt packages to version 0.2.0 and introduces breaking API and storage changes for embedded PostgreSQL. WASIX variants remove temporary and application-data storage variants, and browser IndexedDB uses a new per-database v3 layout without migrating prior generations. It also adds SDK unification for embedded PostgreSQL public APIs and transitions wasix-ts host runtimes to Rust Node-API.

1 commit · 84 files

Updated release tooling to validate that local workspace dependencies use either an unrestricted (*) version or an expected normalized local version, while still requiring local path and correct optionality. Wasix contract checks now validate runtime/tooling/ICU dependencies against the resolved runtime version, and target dependencies follow the same rule. Added tests to confirm source and normalized local Cargo version pins are accepted and mismatched pins are rejected, improving correctness.

1 commit · 2 files

This release bumps versions across native, broker, WASIX, and multiple SDKs while unifying embedded PostgreSQL public APIs. WASIX removes temporary or application-data storage variants and introduces a new per-database v3 IndexedDB layout. It also shifts WASIX TypeScript to route host runtimes through Rust Node-API, introduces related lifecycle changes, and performs release and dependency refactoring.

1 commit · 84 files

The release bumps versions across the embedded PostgreSQL runtimes and SDKs to 0.2.0, including native, WASIX, Node-API, and multiple language bindings. It unifies embedded PostgreSQL public APIs across platforms and aligns native and WASIX runtime behavior. WASIX removes temporary/application-data storage variants and uses a new per-database IndexedDB v3 layout without migrating prior generations, with some changes marked as breaking.

1 commit · 84 files

Release commit verification was extended to treat Cargo dependency version requirements of "*" as valid when they point to the referenced local package. Tests were added to ensure only the canonical local version, path, and features are accepted, and that other changes are rejected.

1 commit · 2 files

This release bumps versions across native, WASIX, broker, and multiple SDKs to v0.2.0 and introduces unified embedded PostgreSQL public APIs. It also changes WASIX and browser storage behavior by removing temporary/application-data variants and using a new per-database v3 layout. The release further aligns WASIX TypeScript/Node-API integration and streamlines related release dependency modeling.

1 commit · 84 files

The repository adds and restructures the build, release, and packaging toolchain to improve determinism and product compatibility management. It updates WASIX and extension packaging flows, introduces shared fixture handling, and refines runtime preflight/smoke tooling. CI workflows and release metadata checks are adjusted to align with the new product graph and independently versioned extension/runtime contracts, aiming to make releases more consistent across targets.

29 commits · 465 files

The system forwards visitor country and region to C15t by rewriting consent API requests at both the frontend proxy and backend router middleware. C15t responses are forced to private, no-store and router tests verify requests are not shared via cached responses. Google Analytics consent bootstrap now preserves a measurement rejection made while the C15t policy request is still pending.

4 commits · 20 files

The site’s resume data, JSON resume, and Typst/PDF outputs were updated to show the correct engineering titles, a unified skills section, and a simplified Mumbai location statement. Resume generation scripts and pages were adjusted to remove separate location/mobility handling. Tests were expanded to ensure the revised titles, visible skills, and selected experience highlights remain consistent across exports and don’t include removed phrases.

2 commits · 11 files

The site and résumé content were rewritten to present a senior full-stack engineering positioning with concrete Namefi and Memorang contributions, including AI workflow and shipped CMS/media recommender details. Identity, skills, and structured exports were aligned across homepage, PDF, JSON résumé, structured data, and the machine-readable llms profiles. Automated tests were added/updated to ensure exports stay consistent and avoid leaking internal editing notes.

1 commit · 18 files

The system now supports Namefi AI Credit purchases, balances, ledger history, and refunds using atomic PostgreSQL transactions. It standardizes billing decisions and units, adds backend services and Temporal fulfillment to charge or refund exactly when work is confirmed or cancelled, and exposes admin routes for read-only balance and history. Frontend adds an AI-credits UI surface and cart integration with exact pricing and purchase controls.

1 commit · 29 files

The system adds and verifies AI-credit accounting and refund behavior across lead generation, studio generation, and order fulfillment. It introduces AI credit order processing that can cancel or fulfill AI-credit items transactionally, updates ledger queries with pagination and user scoping, and tightens input/output contracts and schemas. The frontend and shared pricing/context serialization are updated to match the new models, enabling consistent AI-credit display and purchase flows.

1 commit · 33 files

The app now supports an authenticated AI Credits experience, including a dedicated page for viewing balance, purchasing credits, and browsing credit history with better loading and error states. Cart handling and navigation were updated to include AI Credits alongside domains, and order detail/timeline rendering now recognizes AI-credit-only and mixed orders for clearer fulfillment progress.

1 commit · 122 files

The backend now supports Namefi AI Credits as a priced credit system with a ledger and purchasable credit items. Studio generation and leadgen workflows charge credits before executing work and refund on failure, while an order workflow fulfills AI credit purchases atomically. New AI credits contracts expose account balance and ledger history, and the order router accepts credit-only cart purchases alongside domain items.

1 commit · 38 files

The backend admin AI credits router has been updated to replace award listing with ledger history listing, including pagination and richer search across ledger and related order data. The contract schema now exposes getAccount and listHistory only, with ledger-based row types and pagination fields. The admin permission set also removes write access for AI credits, aligning the API to read-only operations.

1 commit · 4 files

Replaced the admin UI for manually awarding AI credits with a new read-focused experience. Admins can select a user to view their authoritative AI credit balance and last activity, and browse an immutable credit ledger with filtering, paging, and mobile-friendly rows. The AI Credits section access was narrowed to reading AI credits, and the page now includes a refresh action to reload ledger data.

1 commit · 7 files

The system now supports tracking per-user Namefi AI credits through a balance account and immutable ledger entries. It introduces cart items for selected credit amounts, order items for purchased quantities, and enforces valid credit amounts and reference relationships. It also removes the older credit awards table and wires the new schema into the database client and exports.

1 commit · 10 files

The repository adds typed validation schemas for Namefi AI credit amounts, ledger entry and billable action types, and structured AI pricing by model and outbound effort. It also introduces schemas for token usage entries with optional detailed breakdowns and per-stage metadata. New helper functions compute generation and outbound prices in USD-cent credits using the validated pricing data.

1 commit · 3 files

The system now allows up to 3,000 GitHub summary requests per UTC month instead of 300. Documentation, configuration constants, and automated tests were updated to reflect the higher monthly budget and its associated stop-boundary behavior. Pricing-based hard maximum calculations and the stored usage expectations were also adjusted accordingly.

1 commit · 4 files

The site now includes a dedicated Work Log route with its own metadata and a daily activity feed. The GitHub activity timeline has been improved with a preview mode, “show more/show less” collapsible behavior, and tighter layout adjustments for rows and days. Navigation and header behavior were updated so the Work Log route is treated as a first-class page.

1 commit · 11 files