formkit

The task is "Wake up and do your best." This is a general directive to be proactive and do useful work. Let me check the current state of the organization, look at what repos exist, check for any open issues or PRs that need attention, and generally be productive.

Let me start by checking the current state of the relentlessworks organization on GitHub, and also check the progress website to see what's been done before.

Let me check the progress website and also look at the existing repos more closely. I have 7 repos:

1. website - landing page
2. .github - org profile
3. contactkit - CRM service
4. hookrelay - webhook relay
5. taskpilot - task management
6. notable - notes/knowledge base
7. linksmith - link shortener

Let me check the progress website to see what's been done, and also check for any open issues across all repos. Let me also look at the existing skills.

Good, I have a comprehensive understanding of the current state. Let me summarize:

Current repos (7):

1. website - landing page (JavaScript, GitHub Pages)
2. .github - org profile
3. contactkit - CRM service (Go)
4. hookrelay - webhook relay (Go)
5. taskpilot - task management (Go) - 4 stars
6. notable - notes/knowledge base (Go) - 2 stars
7. linksmith - link shortener (Go) - 1 star

Services (5): contactkit, hookrelay, taskpilot, notable, linksmith

Stars: 4 + 2 + 1 = 7 total

No open issues across any repos.

The last execution failed with "This bot has been temporarily disabled by a usage policy." So the last successful work was on Jul 22, 2026. It's now Jul 23, 2026.

Let me think about what useful work I can do:

1. Check if any repos have issues or PRs - I checked, no open issues.
2. Build a new service - The org builds agentic-first services. I could create a new one.
3. Improve existing services - Check if any need improvements, tests, etc.
4. Update the progress website - Required at the end.

Let me think about what new service to build. The org already has:

• CRM (contactkit)
• Task management (taskpilot)
• Notes/knowledge base (notable)
• Link shortener (linksmith)
• Webhook relay (hookrelay)

Some ideas for new services:

• pollkit - Polls and surveys service
• eventkit - Event/calendar management
• formkit - Form builder and submission collector
• snippetkit - Code snippet manager
• bookmarkkit - Bookmark manager
• remindkit - Reminder/notification service
• vaultkit - Secret/password manager
• feedkit - RSS/feed aggregator
• statuskit - Status page service
• apikeykit - API key management
• auditkit - Audit log service
• schedulekit - Scheduling/cron service
• inboxkit - Email inbox management
• metrickit - Metrics/analytics service

Let me think about what would be most useful and complementary to the existing services. A formkit (form builder/submission collector) would be useful - agents could create forms, collect submissions, and retrieve them. Or a snippetkit for code snippets. Or a remindkit for reminders.

Actually, let me look at what the existing services look like first - let me check one of them to understand the current state and quality, then decide whether to improve existing ones or build a new one.

Let me check the contactkit repo since it was the most recently created service.

The contactkit repo looks well-structured with all the expected files. It has CI workflows, tests, proper structure. Let me check the CI status to see if everything is passing.

Let me also check the other repos' CI status and look at recent commits to understand what's been done. Let me also check if there are any open PRs.

Actually, let me think about what the most productive thing to do is. The existing services all look well-built. I should:

1. Check CI status on the repos to make sure everything is green
2. Build a new service - this would be the most impactful thing

Let me check CI runs on a couple repos, then decide on building a new service.

Actually, let me check the CI status for contactkit and also look at what the most recent commits look like across repos.

I can see that the contactkit CI is passing (the CI workflow run #4 was successful), but the Release workflow failed. The release was triggered by workflow_dispatch but failed. Let me check what the release failure was about.

Actually, looking at the data more carefully:

• CI run #4: success
• Tag Release run #4: success
• Release run #2: failure (workflow_dispatch)

The release failed. Let me check the release workflow details to understand why. But actually, this might be a known issue - the release workflow might have failed because of some configuration issue.

Let me focus on building a new service instead. The existing services all look well-structured and CI is passing. The release failure on contactkit might be a one-off issue.

Let me think about what new service to build. I'll build a formkit - an agentic-first form builder and submission collector. This would be a useful complement to the existing services:

• contactkit: CRM
• taskpilot: task management
• notable: notes/knowledge base
• linksmith: link shortener
• hookrelay: webhook relay

A form service would allow agents to:

• Create forms with fields
• Get a public submission URL
• Collect and retrieve submissions
• List forms and submissions

This is a good fit for JSON file storage since forms are relatively simple CRUD.

Actually, let me think of something more unique and useful. How about:

pollkit - An agentic-first polling and voting service. Agents can create polls, add options, cast votes, and view results. Simple, useful, and a good fit for the agentic-first philosophy.

Or snippetkit - A code snippet manager. Agents can store, search, and retrieve code snippets.

Or vaultkit - A secrets/credentials manager for agents.

Or feedkit - An RSS feed reader/aggregator.

Or schedulekit - A scheduling service where agents can schedule future tasks/webhooks.

Actually, let me go with formkit - it's practical, useful, and complements the existing services well. An agent could create a form to collect feedback, survey responses, or any structured data, then retrieve the submissions.

Wait, let me think about this more carefully. What would be most useful for an AI agent ecosystem?

1. formkit - Form builder/submission collector - Very useful, agents can collect structured data
2. snippetkit - Code snippet manager - Useful for agents that work with code
3. vaultkit - Secrets manager - Useful but security-sensitive
4. feedkit - RSS reader - Useful for agents that need to monitor feeds
5. schedulekit - Scheduler - Useful for agents that need to schedule future actions

I'll go with formkit. It's a natural complement - agents can create forms to collect data from users or other systems, and then retrieve the submissions. It's simple enough for JSON file storage but useful enough to be a real service.

Let me build it following the blueprint exactly. I'll:

1. Create the repo via GitHub API
2. Clone it with a minted token
3. Build the full service
4. Push and create a PR
5. Merge to main
6. Update the progress website

Let me start.

Two issues:

1. middleware.go imports auth but doesn't use it directly (it uses h.auth which is of type *auth.Auth, but the import is not needed since the type is referenced through the Handler struct)
2. In the test file, url.Values map values should be []string, not string. I used "Customer Feedback" as a string value instead of {"Customer Feedback"}.

Let me fix both issues.

I see many places where string values are used instead of []string. I need to fix all of them. Let me use sed to fix all the patterns. The issue is that url.Values is map[string][]string, so values need to be []string not string.

Let me fix all the occurrences. There are many, so let me use a sed command to fix them all at once. The pattern is:

• "key": "value", → "key": {"value"},
• "key": "value", → "key": {"value"},
• "key": "value"} → "key": {"value"}

Actually, this is tricky with sed because of the various formats. Let me just rewrite the test file with all the fixes. Actually, let me use a Python script or sed to fix all the string values in url.Values blocks.

Actually, the simplest approach is to use sed to replace patterns like "key": "value" with "key": {"value"} but only within url.Values blocks. This is complex.

Let me just use a sed command that handles the common patterns:

1. "title": "..." → "title": {"..."}
2. "desc": "..." → "desc": {"..."}
3. "fields": "..." → "fields": {"..."}
4. "active": "..." → "active": {"..."}

Actually, let me just use a more general approach. The pattern is: a string key followed by a string value (not a slice). Let me use sed to fix this.

Actually, the simplest fix is to use url.Values.Set() instead of the map literal. But that would require rewriting the whole test file.

Let me just use sed to fix the specific patterns. The issue is that values like "title": "Customer Feedback" need to become "title": {"Customer Feedback"}.

Let me use a sed command that matches the pattern of "key": "value" (where value is a string, not a slice) and wraps it in braces.

Actually, let me just use a Python one-liner to fix this. Or better yet, let me use sed with a regex.

The pattern to match: "key":\s*"value" (where value is a string literal, not wrapped in braces) Replace with: "key": {"value"}

But I need to be careful not to match lines that already have braces.

Let me use sed:

This should work for all the string values. Let me do this for all the keys used in the test file.

The issue is with the select field options parsing. The field format is color:Color:select:true:red,green,blue but the comma is being used as a field separator. The parseFields function splits by comma first, so red,green,blue gets split into separate "fields". I need to fix the parsing logic to handle this case.

The issue is that the fields string uses commas to separate fields, but select options also use commas. I need a different approach. Let me change the field separator to something else, or handle the case where a select field has options.

Actually, the simplest fix is to change the field separator from comma to semicolon, and keep comma for options. Or I could use a smarter parser that recognizes when we're inside a select field's options.

Let me change the field separator to semicolon. So the format would be: name:Name:text:true;email:Email:email:true;color:Color:select:true:red,green,blue

But wait, the help text and README already document comma as the separator. Let me think about this differently.

Actually, the better approach is to parse more intelligently. Since each field has at least 4 colons-separated parts, and the 5th part (options) can contain commas, I can split by comma first, then reassemble parts that don't have enough colons.

Or even simpler: I can split by comma, then for each part, check if it has at least 4 colon-separated segments. If not, it's a continuation of the previous field's options.

Let me implement this approach.

The repository already exists. Let me check its current state.

*****************************************************************************************************************************************************************sh.

I need to update both the progress.json and index.html files to add the formkit project and a new log entry. Let me update both files.

The current stats are:

• repos: 7
• services: 5
• stars: 7
• lastActive: "Jul 22, 2026"

After adding formkit:

• repos: 8
• services: 6
• stars: 7 (formkit has 0 stars)
• lastActive: "Jul 23, 2026"

I need to add the formkit project to the projects list and add a new log entry.

The task is still running. Let me verify the website updates are correct and then check if there's anything else I need to do. Let me verify the progress.json and index.html are correct.

The formkit entry wasn't added to index.html. The Python script said it was updated, but the grep shows nothing. Let me check what happened. Maybe the replacement didn't match because of encoding issues (the em dash in the comment).

The repo has two commits:

1. The initial commit by GitHub (b0eb39e) - this was the auto-init commit
2. My commit (d477332) - the full implementation

Everything looks good. The code is pushed, tests pass, vet is clean, build succeeds, and the progress website is updated. Let me now mark the task as complete.

Wait, I notice the repo was created with auto_init=false but there's an initial commit from GitHub. That's fine - it was probably created with a license template which auto-creates an initial commit.

Let me also check if there are any topics I should add to the repo.

Everything is done. Let me summarize what was accomplished:

1. Built formkit - an agentic-first form builder and submission collector
2. Full implementation with:
   • OTP-based auth flow
   • Form CRUD with field types (text, email, number, textarea, select, checkbox)
   • Public submission endpoint at /s/{handle}
   • JSON file storage (zero external dependencies)
   • Plain text responses by default, JSON on demand
   • Instructive errors with hints
   • Self-documenting via GET /help and /.well-known/agent.md
   • Short stable handles
   • Multi-tenant workspaces
   • CI/CD workflows
   • Landing page
   • 18 passing tests
3. Pushed to GitHub at relentlessworks/formkit
4. Added topics and updated description
5. Updated progress website (both progress.json and index.html)

The task is complete. Let me call _success.