Blog
News, launch notes, guides, and comparisons for builders exploring reusable AI agent skills, trust signals, and ecosystem shifts.
Latest posts
Follow product updates, editorial pieces, and practical explainers that help users navigate the marketplace and adjacent agent ecosystems.
-
Audit GitHub Actions workflows for insecure permissions and unpinned actions
This ASE skill uses zizmor to audit GitHub Actions workflows and composite actions for security mistakes before they ship. An agent can scan local repos or remote GitHub repositories, flag risky permission scopes and unsafe workflow patterns, and return plain output, GitHub-native findings, or SARIF for follow-up automation.
-
Post linter and analyzer findings back into pull requests
This ASE skill uses reviewdog to turn linter and analyzer output into diff-aware pull request feedback. An agent can run existing checks, filter findings to the changed lines, and publish inline review comments or annotations instead of dumping raw logs into CI.
-
Fix failing pull requests by analyzing CI errors and pushing targeted repairs
Use GitHub Next’s pr-fix workflow when a pull request is blocked on failing checks and the likely repair is machine-doable. The agent inspects CI failures, traces the root cause, applies a focused fix on the PR branch, and leaves the result in reviewable Git history.
-
Update repository docs automatically after code changes and open review PRs
Use GitHub Next’s update-docs workflow when code changes should trigger documentation maintenance instead of waiting for humans to remember it later. The agent analyzes what changed, generates the missing docs updates, and opens a reviewable PR rather than silently rewriting the docs in place.
-
Convert browser HAR captures into reusable k6 load tests with har-to-k6
Use har-to-k6 when an agent has recorded browser traffic and needs to turn it into a repeatable k6 script instead of hand-writing one from scratch. The skill is about transforming captured sessions into a starter load-test artifact with validation and export, not about listing k6 or Grafana as products.
-
Review visual regression diffs and publish snapshot baselines in CI with reg-suit
Use reg-suit when an agent needs to compare screenshot outputs against a stored baseline and publish a human-reviewable diff report. The skill is for visual regression review loops, not for general frontend tooling, because the agent’s job is to manage the compare, publish, and notify cycle around image snapshots.
-
Prevent broken GitHub Actions workflows before CI runs with actionlint
Use actionlint when an agent needs to inspect GitHub Actions workflow files before a push or pull request lands. The skill checks syntax, expressions, action inputs, runner labels, cron patterns, and a few security footguns so the agent can stop bad workflow changes before CI burns time.
-
Record and replay HTTP fixtures for deterministic Python tests with VCR.py
Use VCR.py when an agent needs to turn flaky, slow, or rate-limited Python tests into stable runs by recording real HTTP interactions once and replaying them from cassette files. The agent decides which requests belong in fixtures, refreshes stale cassettes when upstream APIs change, and keeps external traffic out of the repeat test loop.
-
Simulate network failures in integration tests with Toxiproxy
Use Toxiproxy when an agent needs to inject latency, disconnects, bandwidth limits, or packet-like failure modes into real service calls during development, CI, or incident reproduction. The agent routes app traffic through controlled TCP proxies, applies toxics at the right moment, and reports which dependency paths fail gracefully versus which ones crack under stress.
-
Broken Link Verification for Static Sites and Documentation
Uses htmltest to crawl generated documentation or static site output, detect broken internal and external links, and return a link-focused validation report before deploy. This is a narrow docs QA skill for agents validating already-built HTML, not a generic site generator or crawler listing.