Free freelance leads, cleaned up and ranked live
47,243 leads from 6 live sources, updated every 5 minutes so you can inspect signal quality before committing.
How We Took a New SaaS From Search Confusion to a Clean Google Entity Before Launch
We’re launching KeepCard on Product Hunt tomorrow. Before launch, we ran into a problem that a lot...
[agigante80/actual-mcp-server] fix(transfers): actual_transfers_create returns null from_id/to_id after 73fa670
## Summary `actual_transfers_create` always returns `from_id: null, to_id: null` despite the fix in 73fa670 that attempted to read back the transaction IDs after the write. ## Root Cause The read-back in 73fa670 calls `rawGetTransactions` **within the same `queueWriteOperation` session**, immediately after `rawAddTransactions(..., { runTransfers: true })`. Even though the Actual Budget API guarantees that `transfer_id` is set synchronously (via `updateTransaction$3` inside `batchMessages`), t
[antiwork/gumroad] Sales page email/search filter is not preserved when returning from sale detail
## What On `/customers`, typing a customer email (or any search term) into the search field filters the sales list. Clicking a sale row opens `/customers/:id`. Clicking the `←` "Back to customers" button returns the user to `/customers` with the email/search filter cleared — even though other filters added in #4662 (price, date, country, products, active-only, sort, pagination) are preserved. **Current behavior** - Typed email filter is lost on back navigation. - The Search popover reopens em
[kagura-ai/memory-cloud] bug(sleep): gpt-5-nano returns empty edges due to reasoning token exhaustion — add reasoning_effort=minimal
Follow-up to #421 / #424. After PRs #423 and #425 deployed (`max_completion_tokens` + `temperature` fixes), sleep LLM calls now succeed at the API layer but **all return empty edges**, so `edges_created`, `llm_accepted`, and `llm_rejected` stay at 0. ## Symptom Latest manual sleep run on `kagura-dev` (post-#425 deploy) — report `8b6a08e0`: ``` llm_calls_made: 12 llm_tokens_used: 22,814 ← LLM is being called and consuming tokens llm_call_failures: 0 ← API succeeds (#421 + #4
[kychee-com/run402] run402 tier --help says prototype is 'free/testnet, 7d' but server charges $0.10 USDC
## TL;DR `run402 tier --help` advertises prototype as "free/testnet, 7d", but the server charges $0.10 USDC to subscribe/renew the prototype tier on the x402 rail. Either the help is stale or the server is charging for something that's supposed to be free. ## Repro (v1.40.2) ``` $ run402 tier --help run402 tier — Manage your Run402 tier subscription … Tiers: prototype (free/testnet, 7d), hobby ($5/30d), team ($20/30d) ``` Then: ``` $ run402 tier set prototype {"status":"error","http":402,"e
[kychee-com/run402] run402 allowance status reports funded:true when billing balance is $0 (field name is misleading)
## TL;DR `run402 allowance status` returns `"funded": true` as long as `~/.config/run402/allowance.json` exists — even when the run402 billing balance is $0 and on-chain mainnet USDC is too low to cover any operation. The field can't be used as a pre-flight "can this account pay for things" check, and its name implies the opposite of what it actually means. ## Repro (v1.40.2) ``` $ run402 allowance status { "status": "ok", "address": "0x…", "created": "2026-03-12T…", "funded": true,
[mauriceboe/TREK] [BUG] Budget categories don't appear in booking dropdown until the Budget page has been visited
### Pre-flight checklist - [x] I have searched [existing issues](https://github.com/mauriceboe/TREK/issues) and this bug has not been reported yet - [x] I am running the latest available version of TREK - [x] I have read the [Troubleshooting guide](https://github.com/mauriceboe/TREK/wiki/Troubleshooting) and my issue is not covered there ### TREK version v3.0.7 ### Describe the bug When the app is first loaded, the "Budget category" dropdown on the booking dialog only shows the "Auto (from
I Built a 4.75 Faster Qwen 2.5 Engine for a $200 GPU – Here’s How
My RTX 3050 laptop GPU was crawling at 30 tokens per second with Qwen 2.5‑0.5B. So I tore apart the...
[robdennis/ponzare] [gap-spec] CR-613.4d-0 switch P/T applies after 7b/c in layer order
## Gap **Gap ID:** `613.4d-0` **Maturity:** **architecturally incomplete.** The engine implements layer-7b (set) and layer-7c (modify/pump) continuous-effect iterations in `getEffectivePowerToughness` (`engine/src/cards/AbilityRegistry.ts`), plus inline +1/+1 / -1/-1 counter handling labelled \"Layer 7d\" in code comments. It has **no switch semantics**. The `ContinuousEffect` schema at `engine/src/engine/GameState.ts:580` reserves `sublayer: 'd' | 'e'` but `'d'` is not consumed anywhere in th
[daniellim922/One-Laundry-Project-Alpha-v1] PRD: Refactor seed data — Public Holidays, Local FT timesheet filtering, and partial seed command
## Problem Statement The current seed dataset has three gaps: 1. **No public holiday master data** — `publicHolidays` and `publicHolidayPay` are hardcoded to `0` in payroll vouchers, even though the app now has a public-holiday management feature. 2. **No lightweight production seed** — The only seed option (`db:seed`) generates 9 months of timesheets, payrolls, and advances. There is no way to bootstrap a clean environment with just workers and public holidays (e.g., for production or fresh de
[vorsangerm-a11y/Test-Today] test(profile): cover bio read/write
## Mault Atomic Task **Zone:** `backend/tests/test_profile.py` **Tests:** `backend/tests/test_profile.py` (self — this task creates the test file) **Max Files:** 1 | **Max LOC:** 60 delta **Dependencies:** #75 **Workstation:** mault-ws-dee4 <!-- MAULT-META: structured block for runtime hooks and review agent --> ```json { "seq": 2, "zone": ["backend/tests/test_profile.py"], "deps": [75], "budget": { "maxFiles": 1, "maxLOC": 60 }, "blastRadius": 0, "hubFiles": [] } ``` ## Descrip
[vorsangerm-a11y/Test-Today] feat(profile): render bio on profile page
## Mault Atomic Task **Zone:** `frontend/src/profile.ts` **Tests:** `frontend/tests/profile.test.ts` (created in a later task) **Max Files:** 1 | **Max LOC:** 25 delta **Dependencies:** #74 **Workstation:** mault-ws-dee4 <!-- MAULT-META: structured block for runtime hooks and review agent --> ```json { "seq": 1, "zone": ["frontend/src/profile.ts"], "deps": [74], "budget": { "maxFiles": 1, "maxLOC": 25 }, "blastRadius": 0, "hubFiles": [] } ``` ## Description Wire the bio textare
[vorsangerm-a11y/Test-Today] test(profile): cover bio render and save
## Mault Atomic Task **Zone:** `frontend/tests/profile.test.ts` **Tests:** `frontend/tests/profile.test.ts` (self — this task creates the test file) **Max Files:** 1 | **Max LOC:** 45 delta **Dependencies:** #76 **Workstation:** mault-ws-dee4 <!-- MAULT-META: structured block for runtime hooks and review agent --> ```json { "seq": 2, "zone": ["frontend/tests/profile.test.ts"], "deps": [76], "budget": { "maxFiles": 1, "maxLOC": 45 }, "blastRadius": 0, "hubFiles": [] } ``` ## Desc
[vorsangerm-a11y/Test-Today] feat(profile): return and persist bio on /profile
## Mault Atomic Task **Zone:** `backend/app.py` **Tests:** `backend/tests/test_profile.py` (created in a later task) **Max Files:** 1 | **Max LOC:** 20 delta **Dependencies:** #73 **Workstation:** mault-ws-dee4 <!-- MAULT-META: structured block for runtime hooks and review agent --> ```json { "seq": 1, "zone": ["backend/app.py"], "deps": [73], "budget": { "maxFiles": 1, "maxLOC": 20 }, "blastRadius": 0, "hubFiles": [] } ``` ## Description Update `GET /profile` to include `bio`
[spree/spree_paypal_checkout] Duplicate gateway_payment_profile_id can break PayPal capture finalization
The app/services/spree_paypal_checkout/create_source.rb tries to find or create a payment source: https://github.com/spree/spree_paypal_checkout/blob/b428104b2bfead8506fddf449f887ae479096218/app/services/spree_paypal_checkout/create_source.rb#L38-L42 If there is already a payment source with the same profile with a different user or nil it tries to create a new one and fails doing so due to the uniqueness validation on PaymentSource: ```ruby validates_uniqueness_of :gateway_payment_profile_id
[SalesforceFoundation/NPSP] RD2: Duplicate Installment Opportunity Created When RD and Opportunity Are Inserted via Salesforce Composite API (Cross-Transaction Race Condition)
## Issue When creating a Contact → Recurring Donation → Opportunity → Payment using the Salesforce Composite API, NPSP creates **2 duplicate installment Opportunities** for the next month instead of 1. ## Related Issues This is a regression/variant of: - #5597 — Enhanced Recurring Donation Generates 2 Installment Opportunities on Create (fixed in 3.180) - #6408 — Duplicate Pledged Opportunities When Opp and RD Updated in Same Transaction (fixed in 3.200) Both previous fixes added `alreadyE
[15273/SkiPlannerAI] research(resorts): crowd levels, peak times, and skier reviews data sources
## Goal Find data sources for crowd intelligence and skier sentiment per resort: when is it busy, when is it quiet, and what do real skiers say. ## Data sources to evaluate ### Crowd & conditions | Source | What it provides | |--------|-----------------| | **Google Popular Times** (unofficial) | Hourly busyness via Places API — requires API key | | **Ski resort official APIs** | Some resorts publish live lift queue times (e.g. Verbier, Zermatt) | | **SkiDB / Snowpilot** | Historical snowpack
[arielb1-sun-security/copilot-studio-test] LiteLLM issue summary - 2026-04-23
# Open Issues from berriai/litellm Repository **Retrieved:** 2026-04-23 16:00 UTC **Total Issues:** 30 --- ## Issue List 1. **#26331** - feat(openai): add gpt-image-2 to model prices https://github.com/BerriAI/litellm/pull/26331 2. **#26329** - fix(proxy): avoid blocking event loop by killing engine instead of disconnecting https://github.com/BerriAI/litellm/pull/26329 3. **#26327** - fix(responses): normalize bridged object field https://github.com/BerriAI/litellm/pull/26
[AztecProtocol/aztec-nr] public_dispatch bytecode on v4.2.0 pushes a production contract over MAX_PACKED_PUBLIC_BYTECODE_SIZE_IN_FIELDS = 3000 — is this the intended budget, and if so what's the recommended public-surface ceiling?
Hey Aztec team! :) I ran into an issue where migrating from 4.1.2 to 4.2.0 ballooned my public byte code size with a 74% increase that put me over the total byte size limit, so I had claude run a report on what changed. My main question is, is this the intended final state of >= 4.2.0 and should I work on reducing my public function footprint, or should I hold off on migrating and can expect optimizations in the future? -- Report ## Summary Migrating a production-target contract from aztec-
[gohugoio/hugo] set file creation and modification times, based on front mattter .Date and .Lastmod
file create and modification times play a important role for http content management. * 304 response with if-not-modified-since requests are based on the files mtime * nginx eTag is based on mtime * search engines use it to estimate last meaningfull content change Currently every rebuild generates current file times, fakeing changed files without meaningfull content change and hurting the management of cashes, and search engine the crawler budget. I purpose a build option PreserveFileTimes, i
Want more? Get leads in your inbox ⚡
Free (this page)
- ✅ Top 20 leads
- ✅ Updated every 5 min
- ❌ No email alerts
- ❌ No filters by skill
- ❌ No CSV export
Pro (€9/month)
- ✅ All 47,243 leads
- ✅ Email alerts (instant + daily digest)
- ✅ Filter by skill, source, country
- ✅ CSV export
- ✅ Contact tracking
Join the waitlist — first 50 get Pro free for 14 days.
Why freelancers use HireAlert
Instead of manually checking Reddit, GitHub, Hacker News, and job boards every day, HireAlert aggregates leads from all 7 platforms and uses AI to score each one. A score of 70+ means high intent — the client has budget, urgency, and a clear scope. You respond fast, you win the gig.
How scoring works
Each lead gets a 0-100 score based on: budget mentions, urgency signals, skill specificity, post quality, and source reliability. Leads below 10 are filtered out as noise. Above 50 is worth a look. Above 70 is a "drop everything and respond" opportunity.