Choose the workflow. Verify the work.
Short, specific guidance for buyers and teams choosing SKILL.md packages or shipping AI-built software. Recommendations separate observed evidence from confident claims.
Install agent skills safely with GitHub CLI
Preview the whole package, choose the correct host and scope, pin the reviewed release, preserve provenance, and check updates without applying them.
Use the gh skill workflow → BUYER GUIDE · 7 MINWhich agent skill should I install first?
Match the problem in front of you to a focused free or paid workflow, using real install and purchase evidence from the JustHandled Labs catalog.
Choose a starting skill → SECURITY GUIDE · 9 MINHow to audit a SKILL.md package before installation
Review provenance, instructions, scripts, permissions, network access, outputs, and updates with a practical nine-step checklist.
Use the security checklist → PRE-LAUNCH GUIDE · 11 MINHow to audit an AI-built app before launch
Replace a convincing demo with evidence for completeness, real data paths, failures, authorization, production behavior, and recovery.
Build the release evidence → PAYMENTS GUIDE · 9 MINStripe webhook works locally but not in production
Separate missing events from delivery failures and business-logic errors, then fix live-mode secrets, raw-body handling, retries, idempotency, and subscription state.
Trace the production failure → DATABASE SECURITY GUIDE · 10 MINSupabase RLS audit checklist before launch
Prove table coverage, policy semantics, cross-user denial, elevated-key boundaries, view and storage behavior, and migration state before production traffic arrives.
Build the RLS evidence → TEST QUALITY GUIDE · 10 MINAudit AI-generated tests before trusting coverage
Check assertion execution, expected-value provenance, mocks, snapshots, and deliberate falsification before a passing suite becomes release evidence.
Test the tests → BROWSER TESTING GUIDE · 10 MINHow to test a web app before launch with Playwright
Select the activation and payment journeys that matter, then run them with resilient locators, protected auth state, mobile projects, console capture, screenshots, and traces.
Build the browser evidence → CI TROUBLESHOOTING GUIDE · 9 MINGitHub Actions failed? Find the first real error
Start with the first failing job and step, separate the root error from cascades, check repository context, and verify one bounded fix.
Diagnose the failed run → CODE REVIEW GUIDE · 8 MINWrite a pull request description from the git diff
Choose the correct staged or three-dot diff, separate intent from evidence, report changed files exactly, and keep unrun tests honest.
Build the reviewer map → RELEASE NOTES GUIDE · 8 MINGenerate a changelog from git commits
Choose the exact Git range, preserve both breaking-footer spellings, keep malformed commits visible, and review the SemVer signal before publishing.
Build the release notes → ACCOUNT MAPPING GUIDE · 9 MINHow to identify a sales champion in an account
Separate authority, engagement, and demonstrated influence; label inferred relationships; and choose one evidence-seeking next action per stakeholder.
Build the champion map → SOCIAL PREVIEW GUIDE · 7 MINLinkedIn link preview image not showing?
Check the final Open Graph tags, image URL and dimensions, crawler access, redirects, and cache state in the order that isolates the real failure.
Fix the preview path → QR PAYLOAD GUIDE · 7 MINAudit a QR code payload before printing it
Decode the final candidate, verify its scheme and domain, expose shorteners and credentials, test the intended action, and preserve a fallback.
Run the payload preflight → ENGINEERING PLANNING GUIDE · 10 MINTurn a PRD into a coding-agent implementation plan
Extract the product contract, isolate ambiguities, map dependencies, write observable acceptance criteria, and preview GitHub issues before anything is created.
Build the execution contract → CREATOR STOREFRONT GUIDE · 8 MINMake a Gumroad cover and thumbnail with AI
Use the right cover and square-thumbnail dimensions, constrain text and claims, compare real composition variants, and review the result at storefront size.
Build the cover set → LOCAL DEBUGGING GUIDE · 9 MINApp won't run locally? Check the environment before the code
Preserve the first useful error, then check runtimes, dependencies, environment-variable names, ports, Docker, and required services in a fixed order.
Diagnose the startup failure → SEO GUIDE · 8 MINHow to fix robots.txt and sitemap.xml before indexing
Align crawl rules, sitemap URLs, canonicals, noindex controls, and live responses before asking Google to discover the site.
Run the crawlability preflight → REPOSITORY GUIDE · 10 MINHow to remove commented-out code safely
Classify dead code, preserve comments that carry intent, preview the exact diff, and verify the repository after cleanup.
Clean the comments without guessing → DATA SAFETY GUIDE · 9 MINHow to scan a codebase for PII before sharing it
Inventory schemas, seeds, samples, configuration, logs, and generated artifacts; triage matches without copying sensitive values into a second leak.
Run the personal-data preflight →How these guides are written
Recommendations separate verified marketplace behavior from judgment calls, avoid calling a clean scan proof of safety, and send readers to individual detail pages before any marketplace handoff. Prices and purchase options remain on Agensi, where they are current.
Already know the job?
Search all 98 focused skills by task, category, or platform.
Browse the complete catalog