Four namespaced commands, one shared discipline: every factual claim in every output traces to a specific source you provided.
Get set up
Builds your org profile, scaffolds an empty workspace directory with a README in each subfolder, and generates a personalised data-pack checklist for your next report or proposal.
Find opportunities
Scans RFP listing sites and your priority funders for new calls. Returns a shortlist with go/no-go memos — every claim carries a URL citation.
Report to funders
Turns your CSV/XLSX data, interview transcripts, budget files, and past reports into publication-quality HTML, PDF, and DOCX. Sections lacking source data become visible [NEEDS INPUT] markers — never filler.
Draft proposals
LOI, concept note, or full proposal for any funder or RFP. Mirrors the funder's own vocabulary, structures to their required sections, and gates rendering behind an independent fact-checker.
The one discipline every command shares
- Every factual claim traces to a source — enforced by a schema validator that refuses to render if any claim is unmapped.
- An independent fact-checker reads your source files and verifies claims before rendering, flagging over-claiming and unsupported causal statements.
- Missing data becomes a visible marker, never plausible-sounding filler.
[NEEDS INPUT]is a feature, not a bug. - Consent records are required for participant quotes; a printable consent form template is included.
- Logos are fetched from real sources (Wikipedia, org homepage) or asked for — never fabricated.
What you get out of the box
Publication-quality outputs with hand-tuned typography (Inter for UI, Source Serif 4 for body). HTML with inline SVG charts, print CSS, and self-contained fonts. PDF via the Typst typesetting system, with proper page breaks, footnotes, and native bar / grouped-bar / line / horizontal-bar chart rendering. DOCX built with python-docx so program officers can edit and comment inline.
Under the hood: an XLSX-to-CSV converter that handles merged cells and dates properly, a HEIC-to-JPG script for iPhone photos, an audio transcription wrapper (Whisper), a best-effort logo fetcher (Wikipedia + org homepage + Google favicon), and a scaffold-workspace helper that creates a folder tree with README-per-folder documentation.
Install
In Claude Code:
/plugin marketplace add everyimpactai/ngo-toolkit
/plugin install ngo-toolkit@everyimpactai
System dependencies:
# macOS
brew install typst
pip install python-docx openpyxl
# Linux
sudo snap install typst # or download from typst.app
pip install python-docx openpyxl
Then run /ngo:setup to build your org profile once.
Documentation
→ Full user manual — step-by-step guide from install through daily use. Written for NGO staff, no technical background required.
Support
Bug reports and feature requests live on GitHub Issues. Please do not paste unpublished donor data or beneficiary names in a public Issue — use a minimal fake example that reproduces the bug. For other questions: dev@everyimpact.ai.
See the FAQ first — most install and rendering issues are covered there.