... learning from all the incredible launches, and figured it was finally time to properly step up and say hello! What I'm building: Like many of you, I frequently use web-based tools to edit, merge, or redact PDFs containing sensitive information (like contracts, IDs, and financial documents). Every single time I uploaded a file to a popular cloud utility, I felt an uncomfortable spike of anxiety: Where is this data actually going? Who owns that server? To solve ... ... problem, I built PrivacyScanPDF.com . It s an all-in-one PDF toolkit (Merge, Split, Compress, Sign, and Redact

... real documents as part of their work. Invoices, delivery slips, receipts, statements, forms, anything where you extract fields and then do something with the result afterwards. One real workflow per team, roughly 20 to 100 documents through space ocr. Redacted is completely fine. I set the fields up with you, and we compare the output against your real values together. The number I actually want. When the extraction is wrong, does space ocr notice? Every value it returns is checked ...

... there's no shared webhooks folder they both sit under. so the glob fix doesn't even apply to the three i have today, they were never colocated in the first place. which proves your enumeration vs property point before i even get to a fourth provider. the property i actually care about isn't "is this file in the webhooks directory," it's "does this handler get called by an external party with money attached to the payload ... ... just being awkward. Predicate and redact

... delta, instead of overloading empty fragments to mean something they don't one explicit signal beats three guards you'd maintain forever. On the Anthropic side it's the arrival of the reasoning block that counts, not its contents: redacted blocks are empty by design, so presence is the signal. The SSE test harness counts reasoning reports now, so a spurious one can't hide inside the streamed text. 172 tests , green. Verified live against Ollama (cold and warm start ...

... worker. For UI testing, a remote headed browser is not just a convenience. It changes the debugging loop. Instead of asking "what did the script think happened?", you can ask "what did the user actually see?" ### Harness Code vs Local Agent Harness Another important decision is where the agent capability should come from. One option is to build the entire reasoning layer yourself: call an LLM API directly, manage prompts, retries, tool calls, screenshots, JSON parsing, and safety rules. That ... ... gives maximum control, but it also means the test harness becomes an AI platform project. The alternative

... changing how I think about content strategy. Comment from Samran Elahi(@samran_elahi): @adana  We've built RezunateLLM, which is a middleware that sits between your code and the LLM providers. Two big pieces: PII Guardrails (catching and redacting sensitive data before it hits the model or leaks back out), and Prompt Management (versioning your prompts, rolling back safely, testing changes without redeploying your app). Basically, the layer that keeps LLMs safe. We are launching on Product Hunt soon ...

... that, it can only pull stuff from its knowledge base, stuff that was created by a human mind. But I'm interested in your perspective, which benefits do you see ? Do you have concrete example of letting AI redact threads and produce marketing materials and it felt right, interesting, constructive & worked very well ? NB: English is not my mother tongue and this was redacted

... comments if anyone's curious.) Comment from Abusaid islam said(@abusaid_islam_said): @joan_garcamo  Spot on, Joan! The amount of un-sanitized API logs and DB dumps that get pasted around daily is honestly terrifying. Auto-redacting sensitive patterns (like authorization headers, emails, or secret keys) client-side before processing is a brilliant idea. Added this to the DevHub Toolkit feature list thanks for bringing this up! Comment from Addie Miranda(@addie_miranda): I'm rarely without GitHub ... ... Garcamo(@joan_garcamo): your second question made me think about how causally teams share snippets during debugging. A database sample or API payload can contain sensitive information even after obvious fields are removed. Maybe utility tools should include automatic redaction

... countdown and starts right at the playhead, so it drops in exactly where you want it. - A real multi-clip timeline. Split, move, and trim clips on any track. When you cut or move a clip, its zooms, redactions, callouts, and facecam move and split along with it, so nothing drifts out of place. - The facecam is its own track. Move it, trim it, split it, and delete or restore it. It stays in sync with your screen, even across ... ... gaps and shifted clips. - Sharper overlay editing. Resize redactions

... live PII and signed secrets. So before a captured sequence can become a saved scenario you'd have to scrub it, otherwise "replay forever" also means storing someone's card metadata forever. Are you thinking record-then-sanitize, or redacting right at ingest so the raw never gets stored? Comment from Gal Dayan(@galdayan): right now it's webhook retries on a payment flow that's genuinely kept me up - a provider retries on timeout, we process it twice because ... ... good way to put it, and honestly the right call for v1. baking the precondition into the scenario spec rather than inferring it means the test is deterministic and you can actually explain to someone why it failed, versus

... Following up on the launch thread, few of you asked same good question: transcript is immutable and redaction happens at share boundary, but doesn't local index still hold plaintext secrets at rest, so any agent/MCP client with read access (or synced backup) can pull old key straight out of search? it is a logical deduction. scritty's model is not "your disk is the trust boundary." instead there's opt in at rest encryption on the local store ... ... reads (e.g. corpus wide semantic search) are refused for that token, not silently honored. "one agent asking for everything" only works if you explicitly mint it admin scope. Bottom line is trust boundary is passphrase + token scope, not filesystem. redaction

... Lens is an open-source Rust project that helps developers visualize application, database, and service traffic during local development. It's built in Rust with a transparent proxy architecture and focuses on protocol decoding, request flow visualization, and secret redaction. Lens is actively under development. We're currently looking for contributors interested in: Tokio & async networking HTTP/2 PostgreSQL, Redis & MySQL protocol parsing Ratatui (terminal UI) Integration testing Cross-platform support If you'd like to contribute, the repository ...

... install. You record a task once. It transcribes your narration, captures screenshots at the right moments, and generates a step-by-step guide: SOPs, user guides, acceptance criteria, release notes. The free tier includes full export and PII redaction. I kept hearing the same thing from QA teams, sysadmins, and support engineers: "we want a documentation tool but IT won't approve the extension." So I built what they were asking for. The people who needed it most were ...

... requires no account. Every upload returns a one-time delete link and an optional claim link. If you later sign in with Google, all shares created from your device past and future are collected under a /mine page. Secret redaction happens locally, before anything leaves your machine. The skill removes common credential formats (AWS, GitHub, Slack, Stripe, OpenAI, Anthropic, Google keys, JWTs, PEM blocks, and KEY=, TOKEN=, *_SECRET= environment variables) before upload. The patterns live in your local SKILL.md ... ... your machine; only the content you explicitly choose to share gets uploaded. A caveat worth calling out: links are unguessable but public. Anyone with the URL can read the conversation, and there's currently no password protection. The local redaction

... honest, dealing with PDFs online can be incredibly frustrating. Most free tools bombard you with ads, ruin your file formatting, or worse compromise your sensitive data privacy. I decided to build a cleaner, faster, and more secure alternative. Confileo is an all-in-one platform where you can: Manage PDFs with Zero Hassle: Merge, split, compress, convert, protect, and redact

... wanted a single app that could handle common PDF tasks without sending documents to external servers. Everything is processed directly on the device whenever possible. Current features include: Merge PDFs Split PDFs Compress PDFs Compare Documents Watermark & Redaction Organize Pages Password Protection Smart Workspaces & Tags The app is now available on the App Store: https://apps.apple.com/us/app/pd... I'm actively developing it and would genuinely appreciate any feedback, feature requests, or criticism. What PDF features do you find ...

... building OpenClaw LLM Proxy and would love early feedback. It's a drop-in OpenAI API replacement that scans every prompt and response for PII (SSN, credit cards, API keys) and injection attacks before they reach the provider. Block, redact, or log. We built it because most LLM proxy setups focus on routing and cost - we wanted to add a security layer on top. Self-hosted, 113 tests, Docker one-command. GitHub: https://github.com/secure-mind-l... What would you want ... ... partha_mehta): Update: We're launching on Product Hunt this Tuesday (June 16) at 12:01 AM PDT! Would love your support. Pre-launch page: https://www.producthunt.com/prod... Since posting here, we've added a demo video, comparison table vs

... executed - verdict for the run: matched policy, drifted, or unverifiable The beta failure cases I would want are false-positive kill during a long but valid tool call, policy update between run and review, missing trace segment, and PII redaction hiding the evidence needed to review the stop. I am working on a receipt layer around this same problem and would be happy to cross-test a small external fixture if useful ...

... generated code, tool outputs, repository context, approvals, and policy decisions all tied together. At that point retention is no longer just how long do we keep logs? it becomes a lifecycle governance problem around what can be stored, exported, redacted, deleted, or legally preserved across the entire AI workflow. Comment from Lucy Bennett(@lucy_bennett1): This feels closer to audit infrastructure than a normal AI coding tool, which is probably the right direction. Comment from Muhammad Usman(@datawithusman): Retention first ... ... treated as entirely independent actions either. If a policy blocks a change, the retry path needs provenance continuity: what changed between attempts, whether the prompt/model/context materially shifted, who initiated the retry, and whether the policy conflict was actually resolved versus

... latest beta room previews the future shopify_oauth_states table, including: state fingerprint payload hash signature hash session fingerprint tenant binding workspace binding Shopify store binding operator user binding expiry window consumed-at marker replay tracking callback receipt reference redacted metadata only The goal is to make Shopify OAuth state one-time-use, auditable, replay-resistant, tenant-bound, and safe before enabling any real exchange path. What is still left before first test users: Create the actual OAuth state vault ... ... after state exchange is proven. Token storage is its own separate gate. The system should not persist live Shopify access tokens until: OAuth exchange is working safely state replay protection is working token vault insert path is tested token redaction

... latest beta room previews the future shopify_oauth_states table, including: state fingerprint payload hash signature hash session fingerprint tenant binding workspace binding Shopify store binding operator user binding expiry window consumed-at marker replay tracking callback receipt reference redacted metadata only The goal is to make Shopify OAuth state one-time-use, auditable, replay-resistant, tenant-bound, and safe before enabling any real exchange path. What is still left before first test users: Create the actual OAuth state vault ... ... after state exchange is proven. Token storage is its own separate gate. The system should not persist live Shopify access tokens until: OAuth exchange is working safely state replay protection is working token vault insert path is tested token redaction

... streaming security engine for AI-generated code. Instead of scanning code after it's written, Cencurity runs inline with code generation and enforces security decisions in real time. What it does: - blocks dangerous code (eval, subprocess, unsafe execution) - redacts secrets (API keys, credentials) - enforces policies during generation Why this matters: AI is now writing real production code. But security workflows are still built for post-generation analysis (SAST, reviews, etc). That gap is where insecure code slips through. CAST closes ...

... built PDF Redactor Pro a Windows desktop tool that permanently removes sensitive information from PDF files before sharing. It auto-detects emails, phone numbers, SSNs, credit cards, and more. Supports keyword search, RegEx patterns, image and table redaction, batch processing all 100% offline. Your documents never leave your PC. Right now it's free, or pay what you want https://payhip.com/b/mesxD ...

... everyone , I've been building Prompt Armour a browser extension that detects and redacts sensitive data in real-time inside ChatGPT, Gemini, and Claude before you hit send. The problem: every time you paste code, emails, documents, or customer data into an AI chatbot, that data potentially gets stored, logged, or used for training. Most people don't think twice about it. How it works: - Scans your input in real-time as you type or paste - Detects names, emails, phone ... ... paint highlights without touching the DOM. Built with Plasmo, TypeScript, React. Available on Chrome Web Store now, Firefox version pending review. I'm trying to figure out what to build next. A few things on my radar: - Standalone redaction

... formats ChatGPT alone won t reliably extract from, translation and summarisation + original file preview Secure structured sharing with fine-grained access controls - share all your health picture or only the pieces you choose Personal identifiable information detection and selective redaction, sanitised files export And so much more! I want to keep my story momentum . Check all our current features here I've been hosting "tidy up your data" meet ups to request, find and upload health data. With the current ...