organic-os welcomes bug reports with repros, code fixes with tests, new skills and agents, CMS and channel adapters, docs, and evidence updates backed by primary sources. Business data never lands in the engine repo; CI enforces that boundary on every pull request.
What we welcome
- Bug reports with a repro: what you ran, what you expected, what happened, plus your Python version.
- Code fixes with tests: the core libraries are plain Python covered by the test suite. A fix without a regression test that would have caught it is a partial fix.
- New skills or agents that read the site profile and work for any site, not one business.
- CMS or channel adapters: a new write path alongside the WordPress one, or a new approval channel alongside Telegram, pr-merge, Slack, and email.
- Docs: fixes, clarifications, missing setup steps.
- Evidence updates with primary sources: a change to the evidence table needs a real citation, a study or a controlled experiment, not “I heard AI cites X more now.”
The data boundary, in one paragraph
organic-os is an engine, not a dataset. Your business data (site profile, keywords, brand rules, competitors, skillbook, signals) lives in your private brain repo, never in the engine repo; CI fails any pull request that introduces it. Engineering rule and privacy guarantee in one: the engine stays generic, and nobody’s competitor list is one accidental git add . away from a public PR.
Sharing learnings without sharing data
Found a tactic that works? Open an issue with the tactic, its evidence tier, and the source. No URLs, keywords, brand names, or screenshots of a real site.
Where the project is heading
The roadmap is themes, not promises, and issues against any item are welcome.
- v0.2, first-week wins and loop hardening, complete: schema versioning and migration, gate self-verification, the connector wizard with live verification, cannibalization and content decay detectors, site drift watch, per-run cost transparency, dry-run mode for apply, Bing Webmaster plus IndexNow submission, and approval expiry for stale items.
- v0.3, execution breadth, complete: the CMS adapter contract plus WordPress and git-static adapters, internal-link graph analysis with a gated redirect and 404 workflow, digital-PR mention signals plus entity consistency plus AI-referral segmentation, comparison-content briefs plus topic clustering plus editorial-oversight scoring, and the claude-seo audit import.
- v0.4, adapter breadth and editorial depth, in progress: an analytics adapter slot, an image-generation adapter slot, a gated image and alt-text fix workflow, and an editorial-policy section in the site profile.
- v1.0, many sites, many hands: multi-site orchestration across brain repos, agency mode, a content inventory and portfolio view, read-only Reddit and community observe, and AI-crawler analytics from server logs.