Week of 5/18 - Development Standup Thread

[azdak]

6-Month Roadmap
Hung Scan Postmortem
Dev Report - Period ending March 6th
Dev Report - Period ending March 31st
Dev Report - Period ending April 5th

New/Discussion

• Refactor out all hardcoded paths/regions/names/etc to a centralized env file

• Determine IaC/Provisioning Framework (Terraform, OpenTofu, whatever): there are a variety of options for Infrastructure as Code (IaC) and we need to look at the tradeoffs for each and settle on one

• Create BUILD SCRIPTS 💀: Write IaC scripts to provision all the infrastructure-ie lambdas, DB, and relationships between them- needed to run an Equalify instance

• Determine deployment strategy: Figure out how to handle deployment, ie how code (and code updates) are pushed to the provisioned infrastructure

• Provisioning & Deployment Documentation: Technical documentation for deployment

• User Documentation: Developer-focused Readme to introduce how to use the app from a high level.

• Blockers Refactor We are currently retaining all blocker data, including blocker data from old scans, in the blockers db table, which is causing a performance bottleneck (roughly 20M blockers are currently in the blockers table]. the only place the old data is currently being used is for the blocker counts for old scans.

  • old blockers should instead be moved to a separate table, and the blocker count for a particular scan cached as a number field in the scans table
  • this could either be done in scanwebhook upon receiving results, or as a recurring task (in runseveryminute). In the latter case, we wouldn't need a separate DB migration

Features

• Crawl and WP integration
  • Deploy new stuff on staging
  • RED plugin deployment on Monday
• April Report

Bugs

Pending Next Roadmap Meeting

• Add PDF Link Source Tracking in Equalify #618

Current Work

@azdak To Dos:

• Look into slow blockers table fetching
• Mobile breakpoints
• Fix tables CSS
• Admin-settable co-brand logo
• Scanning state indicator on Audits screen would be nice (@azdak 4/30)
• Incorrect URL counts
• Align language used in the WP plugin & Build screen to ensure they're consistent, and emphasizes the sync functionality
• Investigate strange PDF metrics. The scanStarted metric is reporting very high numbers of PDF scans, and PDF-heavy audits are showing some very high blocker count numbers. We should look into this!
  • ID'ed outstanding issue with missing db constraint triggering cascade of errors saved to the db. @heythisischris currently on it
• Should we add UX Perf metrics?
• Clicking into blocker shows "Blocker not found", but after a few seconds it loads / appears. Perf issue likely.
  • Think we just need to add a loading state, but a little blocked due to the shortId issue
• Improve Blocker filter categories
• Fixed bug with Blocker table content type filtering
• WebAIMee-style LLM for improved individual blocker info
  • Staging Version
  • Testing
  • Move to Prod
    • Add new blocker_llm_summaries table
    • set DB Permissions for new table
    • set AWS bedrock permissions for prod api lambda
• (@azdak 5/8) Add a banner or other visual indicator to differentiate the shared/public read-only version of an audit; currently the distinction is not obvious to users, and this bug took a while to figure out, due to it actually only occurring in the share version. A more verbose/obvious gDocs-like modal after you click the "Share" button is also not a bad idea.
• Stats Package: would be good to have somewhere centralized (maybe a tab under Account) with some general usage stats: users, audits, scans last 30 days, audits created last 30 days, etc

@heythisischris To Dos:

• Look into slow blockers table fetching
• Show total blockers count
• Blockers CSV export only exports current page in table
• Weird issue:
  • Scanner->results webhook can be:
    • OK
    • Error
    • but also Error ... Error ... OK
  • Means Count on errors is tricky/unclear to user: if I scan 50 pages, I could get 40 (or 80!) errors but still have 50 successful scans
  • Options
    • OK overrides/drops errors from table: drops data but simplest/does what it says UX. Is first try error data important?
    • No retry: simplest to understand but uh kinda counter to the goal, which is scanning
    • Retain but only show "final" errors in UX: most complete but not sure easiest way to implement
    • 5/11 - Try tying errors to a blocker ID, use the latest error. Update error depending on what's latest? Or array of errors? We'll figure this out.
• Work on DB Schema version control and Hasura metadata version control
  • Include in CI/CD
• Building Chrome extension prototype
• Short ID namespace collision/Duplicate Blocker ID/"Scans Not Competing"
• Work with Amanda to update docs
• Update docs with new site onboarding workflows (crawl & WP integration)
• Complete May 2026 Report
• Work on semantic releases, add Releases to Hub / We should probably settle on a versioning system/strategy going forward! (Added 4/29, on @heythisischris's list)
  • Something like semantic-versioning might be easiest. Would mean standardizing commit message format
  • Need to document any new system
  • Should also surface current version to UI somewhere
• Investigate detailed view performance, do small audit on scalability

@heythisischris Icebox

• Get clean AWS account provisioned (talk to Dean)
• (proposed 1/22) Fix URL for prod Hasura
• Coordinate meeting w/ Helen & Michael on documentation workflow AND ask about how Equalify communications are handled, what team is responsible, and how we can properly coordinate updates / FAQ / etc.

Future

Current Feature Development Priority Order

Reviewed every 3rd Thursday | Feature Priority Sheet

Task	Tech Notes
Admin-settable co-brand logo	Allow admins to set a co-brand logo to appear alongside Equalify
Improve Blocker filter categories	Right now, we are mapping not very useful data to the Category fields, especially with PDFs. The goal here would be to identify the most useful data to map to categories
Show total blocker count on Audits screen
WebAIMee integration	Either integration with existing external WebAIMee service, or integrating this directly into Equalify. Would add LLM-based "How to Fix" information to individual blocker pages
FOSS/Infrastructure Sprint	Develop build scripts (with documentation) to more easily allow anyone to spin up an Equalify instance
Separating Template vs Content Blockers	Difficult problem, immediate way forward may be through expanding the new WP plugin
Design/Maintenance Sprint	Dashboard visual refresh and general code cleanup
Self-Registration	The idea here would be a public page for SSO installs, from which a user could enter email to request access. Approve/Deny option would then appear on the admin screen

Planned Sprint Details

FOSS Sprint aka Infrastructure Week (1-2 weeks)

• Determine IaC/Provisioning Framework (Terraform, OpenTofu, whatever): there are a variety of options for Infrastructure as Code (IaC) and we need to look at the tradeoffs for each and settle on one
• Create BUILD SCRIPTS 💀: Write IaC scripts to provision all the infrastructure-ie lambdas, DB, and relationships between them- needed to run an Equalify instance
• Determine deployment strategy: Figure out how to handle deployment, ie how code (and code updates) are pushed to the provisioned infrastructure
• Provisioning & Deployment Documentation: Technical documentation for deployment
• User Documentation: Developer-focused Readme to introduce how to use the app from a high level.

Design & Maintenance Sprint (1-2 weeks)

• Get alignment on overall visual direction and palette
• Branding: Defining how branding is handled on the system.
• Design System: Atomic system or Storybook for component uniformity
• Darkmode 😎
• Mobile breakpoints
• Style and Code Cleanup: Finish componentizing, minimize global styles, and code cleanup
• Discussion/Fit into roadmap (@azdak 3/16): Roll-up reporting, eg being able to get a summary view of several audits at once
• Discussion/Fit into roadmap (@azdak 3/16): user<->audit groups and/or more fine-grained control over what each user sees on the audit listing page

Onboard Features Sprint (1-2 weeks)

• Site onboarding features
  • Lambda for site crawling
  • Integrate onboarding features into Build Audit
• Allow external, linked CSVs as basis for Audit URLs (with regular or just-in-time on-scan re-scrapes), allowing WordPress plugin integration
• CSV User import

Backlog/Blocked

• Help text first pass
  • ID icon areas
  • Refactor HelperTooltip component
  • Write help text
• Integrate error reporting into "blockers over time" scans table (also possibly the graph?)
• Discussion/Fit into roadmap (@azdak 3/16): the eternal Custom Rules question
• Discussion/Fit into roadmap (@azdak 3/16): the even more eternal "exclude the CMS header/footer" question 😄

[bbertucc]

Glad build scripts seem to be moving up. I would just highly suggest figuring out how to be less dependant on AWS. I don't know when, but I can safely say that we will need to be Bezos agnostic. Too many community members don't use AWS.

[heythisischris]

Equalify Stale Blocker Analysis: https://docs.google.com/document/d/1duSPv3KCgtMh8OUX1j1dBzJKQabSIW-mVf_qk4Pl1O0/edit?usp=sharing