tiny-runtime 0.1.1 — API pagination, retry loops and tool progress

Safety and efficiency update: incremental crash-recoverable checkpoints, bounded collection callbacks, successful repeated reads allowed, duplicate cache hints coalesced, and expired/late cache acknowledgements rejected. Existing version-1 checkpoints can still resume.

Three MCP tools and a JavaScript/TypeScript SDK by AVRG3 for agent workflows:

• collect_pages: resumable API pagination and local batches, duplicate checks, exact integer totals, explicit complete/partial/failed results.
• check_progress: inspect repeated failures against measured state; SDK wrappers enforce the guard in an agent host.
• plan_cache: tool-progress cache scheduling hints; live GPU effects require a compatible serving-engine integration.

Download the MCPB for an MCPB-compatible client (same file on macOS, Windows and Linux, dependencies included). Requires Node.js 20+ or a compatible host runtime. This exact bundle was tested through MCP on all three operating systems.

For other clients, use the tarball with npx; it downloads dependencies and needs no npm account:

npx -y -p https://github.com/Warddamn/tiny-tools/releases/download/runtime-v0.1.1/tiny-runtime-0.1.1.tgz tiny-runtime-mcp

Install and tool reference · Try three examples

No model calls or telemetry in the tools. HTTP sources are explicit read-only requests. The MCP trace tools do not automatically intercept other clients or control hosted-model caches. No real-world token, compute or GPU savings are claimed. Bundle is unsigned; SHA256SUMS.txt and registry metadata verify artifact integrity.