@promethean-os/report-forge

Create terse, actionable Markdown reports from GitHub issues using a local LLM (Ollama by default).

• Native ESM, functional style, immutable transforms.
• Flat package, .js extension imports in TS.
• Tested with AVA.
• License: GPL-3.0-only.

Install

pnpm -w add -D @promethean-os/report-forge

CLI

report-forge riatzukiza/promethean docs/reports/promethean_issues_report.md

Requires an Ollama server (default http://127.0.0.1:11434) and optional GITHUB_TOKEN for higher rate limits.

API

import { generateReport } from '@promethean-os/report-forge'
import { ollama } from '@promethean-os/report-forge/dist/lib/ollama.js'
import { github } from '@promethean-os/report-forge/dist/lib/github.js'

const gh = github(process.env.GITHUB_TOKEN)
const issues = await gh.listIssues('riatzukiza/promethean', { state: 'all' })
const md = await generateReport({ repo: 'riatzukiza/promethean', issues }, { llm: ollama() })

Philosophy

• Separation of concerns: fetch → summarize → render.
• Pure functions: transform data without mutation.
• Templates last: the LLM polishes content, not control structure.

Tests

pnpm --filter @promethean-os/report-forge test

Config

• Ollama: set OLLAMA_BASE and OLLAMA_MODEL via wrapper if you fork ollama()
• GitHub: set GITHUB_TOKEN