[recipes] Add editorial-policy + weekly auditor recipe#276
[recipes] Add editorial-policy + weekly auditor recipe#276HansBohlmann wants to merge 2 commits into
Conversation
A 40-rule constitution that governs synthesis prompts plus an Edge Function that audits compliance weekly. Trait-fix discipline encoded as a numbered policy + auditor pair, so drift is detectable instead of scattered across prompt strings.
|
Hey @HansBohlmann — welcome to Open Brain Source! 👋 Thanks for submitting your first PR. The automated review will run shortly and check things like metadata, folder structure, and README completeness. If anything needs fixing, the review comment will tell you exactly what. Once the automated checks pass, a human admin will review for quality and clarity. Expect a response within a few days. If you have questions, check out CONTRIBUTING.md or open an issue. |
alanshurafa
added
area: recipes
|
Thanks for the contribution, and welcome. This is a substantial, well-designed recipe — pairing a numbered editorial policy with a weekly compliance auditor is a coherent approach to keeping synthesis output disciplined. Self-contained under The schema change is light: — Alan (community reviewer; non-binding) |
2 participants
Contribution Type
/recipes)/schemas)/dashboards)/integrations)/skills)What does this do?
A
recipes/editorial-policy/contribution: a 40-rule constitution (editorial-policy.md) that governs every synthesis prompt in an OB1 instance, paired with a weekly Edge Function (auditor/) that audits compliance and posts critical findings to Slack.Requirements
pg_cronandpg_netextensions enabledgpt-4o-mini)No canonical skills or primitives required.
Checklist
README.mdwith prerequisites, step-by-step instructions, and expected outcomemetadata.jsonhas all required fieldsWhy
Most OB1 forks fix prompt drift in scattered prompt strings, so the same failure modes — inflation of thin sources, drift across syntheses, smoothed-over contradictions — keep reappearing under new names. This recipe replaces that pattern with a numbered policy that prompts cite (
R3.5,R5.3,R10.2) plus an auditor that enforces compliance. When something drifts, you fix the policy and the rule propagates to every prompt that cites it. When the auditor finds a violation, it cites the rule it broke.What's included
editorial-policy.md— the 40-rule doc, generalised for community adoption (R1.1andR9.3are the operator-specific rules to adapt). Versioned changelog shows the trait-fix disciplineR10.3prescribes.auditor/index.ts+auditor/deno.json— Edge Function that runs weekly, returns structured JSON findings, stores them asaudit_reportthoughts (append-only perR8.1/R8.3), posts to Slack only onseverity:criticalper low-noise discipline.schema.sql—get_recent_audit_reportsRPC + partial index. No new tables.schedule.sql— pg_cron weekly entry with placeholder URL/key.README.md— full setup walkthrough following the_template/pattern: prerequisites, credential tracker, 7 steps, expected outcome, troubleshooting, customisation.Adopting
Total setup time: ~30 minutes. The
README.mdwalks through it. Adopters edit R1.1 (operator name), R9.3 (timezone), and update their existing synthesis prompts to start withFollow Open Brain Editorial Policy v{version}per R10.2.Why a recipe, not a skill or extension
Three things that only deliver value together — a policy doc, an SQL helper, and an Edge Function. The policy without the auditor is documentation that drifts. The auditor without the policy is a function with nothing to check. Recipe pattern fits.
Tested
Running in production on my OB1 instance. The v1.3 changelog entry in the policy doc is a real trait-fix from an actual briefing-inflation incident — a one-line reminder ("X follow up is urgent") got paraphrased into a theme + worth-revisiting reflection + philosophical "prompt for today" across four sections of one morning briefing. Adding R3.5 and rebuilding the synthesis prompts to inherit it fixed it. That's the discipline this recipe encodes.