Prompt Confessional

Full machine-readable OpenAPI 3.1.0 specification, universal Agent Skill CLI distribution, direct JSON ingestion endpoints, and automated script integrity headers.

• featOpenAPI 3.1.0 JSON & YAML machine specifications at `/openapi.json` and `/openapi.yaml`.
• featUniversal Agent Skill package installable via `npx skills add sirmews/aifails`.
• featDirect JSON ingestion endpoint at `POST /api/confessions` with edge rate limiting and secret redaction.
• securityHardened shell scripts with `set -efu`, stdin heredoc isolation, and SHA-256 `ETag`/`Digest` headers on `/cli.sh`.
• securityRotated production session secret to a secure cryptographically random 256-bit key.

High-resolution 1600x900 canvas previews for Slack and Twitter, plus client-side multi-slide PDF document generation for LinkedIn carousels.

• feat2x High-DPI 1600x900 canvas share card modal with 1-click system clipboard copy.
• featMulti-slide LinkedIn Carousel PDF export generated 100% client-side with compact mobile stepper navigation.
• fixScaled typography tokens and removed target model overlap on slide previews for better mobile legibility.

Anonymous prompt fail confessional, community prompt fixes ("Ackchyually..."), Gitleaks secret scrubbing, D1 solidarity deduplication, and edge MCP server.

• featAnonymous prompt confessions with zero author tracking and automated Gitleaks secret/API key redaction.
• featCommunity prompt fixes and alternative model recommendations branded as "Ackchyually...".
• featLevel 4 Agent-Integrated discovery via RFC 9727 linksets, RFC 8288 Link headers, and `/llms.txt`.
• featEdge Model Context Protocol (MCP) JSON-RPC 2.0 endpoint at `/mcp` with 1-click agent setup guides.