Skip to content

Latest commit

 

History

History
27 lines (18 loc) · 2.54 KB

File metadata and controls

27 lines (18 loc) · 2.54 KB

Patch Notes

v0.5.0 — 2026-04-15

increase spot-analyzer character limit from 300 to 500

The spot-analyzer now accepts prompts up to 500 characters — both the backend API validation and the web dashboard textarea, counter display, and warning threshold have been updated to reflect the new limit.

v0.4.0 — 2026-04-15

rename binary to piq and add start/stop daemon commands

The CLI binary is now piq instead of promptiq — all existing commands work unchanged under the new name. Two new commands handle the web dashboard as a background process: piq start spawns a detached server, writes a PID file to ~/.promptiq/serve.pid, and opens the browser automatically (with sudo re-exec for port 80); piq stop sends SIGTERM and cleans up the PID file. Both commands handle stale PID files and double-start gracefully. A migration note in the README covers updating the Claude Code hook and cron entries.

update README with pnpm commands and WSL background server setup

Development commands in the README now reference pnpm instead of npm. Added a "Running in the Background" section documenting the new piq start / piq stop workflow for WSL users who need port 80 with sudo.

v0.3.0 — 2026-04-15

Switched prompt batch serialization and week-day entries to XML tags (<prompt index="N">, <day date="…">), improving scoring consistency and reducing prompt misattribution. Fixed the Analyzer tab's before/after examples and improvedPrompt output to generate natural English instead of mechanical rewrites, and bumped the spot-analyzer token limit to avoid truncation. Also fixed promptiq serve to default to port 80, corrected a regex escape in the dashboard diff view, and resolved a home-directory bug that caused the dashboard to show no history when started with sudo.

v0.2.0 — 2026-04-15

Added on-demand Analyzer tab and Analyze button for single-prompt and batch analysis. Rubric and analyzed prompts are now wrapped in XML tags with escapeXml() escaping, preventing prompt injection from user-authored rubric files.

v0.1.1 — 2026-04-15

Fixed a bug where the Analyze button re-analyzed all of today's prompts on every run. Now only prompts logged since the last analysis are sent to the LLM.

v0.1.0 — 2026-04-14

Initial release. Includes prompt logging, rubric-based daily analysis, DRM weekly/monthly rollup, web dashboard (Status/Patterns/Weekly tabs), on-demand spot analyzer, MCP server, slash-command classifier, tip tracking, and feedback correlation.