HN Showcase
Back to browse
GitHub Repository

AI-powered PR & issue triage for maintainers. Dedup, scoring, vision doc alignment.

9 starsTypeScript

Treliq – PR triage CLI with 20 signals and optional LLM scoring

by chrismagno·Feb 19, 2026·1 point·0 comments
Visit ProjectView on HN

AI Analysis

●●SolidNiche GemSolve My Problem
The Take

Deduping PRs and scoring them with 20 heuristic signals is a concrete, useful idea — especially the scope-coherence signal and embedding auto-fallback for providers without embeddings. The repo supports CLI, a persistent server, GitHub App integration and an explicit --model flag for provider flexibility, but it's still early and adoption/UX examples (ranked output, workflows) are thin — promising engineering scaffolding that needs real-world validation.

Target Audience

Open-source maintainers, repository owners, engineering leads and anyone juggling large PR queues

Post Description

CLI + dashboard for open-source maintainers to decide which PR to review/merge first. 20 heuristic signals (scope coherence, complexity, staleness, CI status, etc.) + optional LLM scoring via Gemini, OpenAI, Anthropic, or OpenRouter. v0.5.1 adds --model flag and embedding auto-fallback. Would love feedback from anyone managing large PR queues.

Similar Projects

Developer Tools●●Solid

Open Slop – A GitHub Action to Triage AI-Generated PR Slop

Instead of chasing unreliable "AI fingerprint" heuristics, this action flags PRs using three blunt but practical signals — Velocity (how fast complex changes appear), Shotgun (many unrelated PRs from the same account), and Ghost (account age). It’s a small, sensible tool you can drop into a repo (bundled dist, single triage comment) that will immediately reduce the noise; just watch for false positives around rapid expert contributors and consider tuning thresholds.

Big BrainShip It
__alberto
333mo ago
Data●●Solid

GoldenMatch – Entity resolution with LLM scoring, 97% F1, no Spark

Fellegi-Sunter matching with active learning beats Dedupe.io on complex datasets.

Big BrainSolve My Problem
benzsevern
302mo ago
AI/ML●●Solid

HiddenState – How I keep up with 500+ ML papers a day

Clustering by the specific technical constraint being attacked — not by topic — and scoring each signal on convergence, implementation evidence, engagement and significance is a neat, high-signal trick for surfacing research trends. It smartly dedupes org noise and ingests many sources, though using Claude as a clustering black box means the scoring pipeline could use clearer auditability or export hooks for skeptical researchers.

Big BrainNiche Gem
CosmoSantoni
113mo ago