sickn33/agentic-awesome-skills v18.7.0

v18.7.0 "Prompt Provenance, LintLang Audits, and a Fully Categorized Catalog"

on GitHub

[18.7.0] - 2026-09-27 - "Prompt Provenance, LintLang Audits, and a Fully Categorized Catalog"

Adds 2 reviewed skills for verifying what a shipped AI product was actually told and for static auditing of agent instructions, refreshes the Jev Social and Unified AI Gateway runtime references, and classifies every remaining catalog entry for 2,474 skills with zero uncategorized records.

A catalog release for Claude Code, Cursor, Codex CLI, Gemini CLI, and related AI coding assistants. It includes the complete protected maintainer batch merged after 18.6.0, with existing installation interfaces preserved.

Start here

• Install: npx agentic-awesome-skills@18.7.0
• Choose your tool
• Bundles
• Workflows

Added

• System Prompt Lookup (#1628) — system-prompt-lookup answers "what is this product actually told?" from a dated archive of captured prompts and tool schemas instead of a fluent recollection. Requires quoting only what was fetched in the session, labels captured versus vendor-reported artifacts, and treats fetched prompts as data rather than instructions.
• LintLang Audit (#1638) — lintlang-audit runs the pinned LintLang 0.8.0 static checks over a named agent instruction file, tool definition, or supported Python prompt, and reports per-file verdicts, severity counts, finding codes, and locations without editing files or calling a model. Read-only, risk: safe, with the Apache-2.0 notice linked at its pinned upstream revision.

Changed

• Jev Social (#1637) — refresh the pinned runtime from v0.1.5 to v0.1.8 (5270e23cfd27aace9055669ee396926973baa241); the read-only browser boundary and explicit-media intent are unchanged.
• Unified AI Gateway (#1627) — revendor the skill so the current v0.8.0 release and its fifteen tool names are distinguished from the digest-pinned, content-reviewed 0.4.9 image that ships nine. The pin and digests are unchanged, and the content-review links stay on the immutable 8561ec5 revision.
• Catalog categorization (#1641) — add a reviewed, source-controlled override map that classifies the 359 remaining uncategorized entries, so the catalog now reports 0 uncategorized skills.
• Release reproducibility (#1642) — derive guide sitemap dates from the last substantive change rather than the canonical-sync commit or the wall clock, so canonical-sync PRs are byte-for-byte reproducible.
• Hosted catalog and documentation (#1629, #1631, #1632, #1633, #1635, #1636) — publish the user guides in the catalog site with full-text search, Git-derived dates, article and breadcrumb structured data, and real sitemap dates; restore the homepage catalog provider; and unify the site theme and typography across routes.
• Release surfaces — regenerate the catalog, offline AAS Core assets, Codex/Claude plugin mirrors, editorial bundles, marketplace manifests, compatibility reports, and web assets for 2,474 skills.

Documentation and community

• Added the hermes-labs-ai/lintlang and Continuum-AI-Corp/OrcaPromptVault source credits to the README.
• Thanks to @xizhuomengcontin for system-prompt-lookup, @roli-lpci and Hermes Labs for lintlang-audit, @FrancoStino for the Unified AI Gateway revendor, and @IRONICBo for the Jev Social runtime refresh.

Validation scope

Exact-head maintainer review for changed skill content, protected source PRs, repository validation and tests, reference validation, docs security, protected CI and CodeQL, dependency review, canonical synchronization, release preflight, Linux/Windows preview checks, npm publication, and release provenance verification.