README.md

# Eastern Orthodox Source Study

Version **0.1.8**. A separate, skills-only redesign based on Eastern Orthodox Theology 0.9.16. It does not replace the approved 1.0.0 submission or connect to an MCP server.

## Purpose

For readers at any level, from everyday questions to detailed research. Precise sources and contextual reading guidance remain the focus.

Ordinary Orthodox questions lead to contextual readings: identified works, inspected passages, useful reading ranges, brief reasons for selection, and material limitations. Dedicated authentication questions receive a bounded finding with evidence the reader can inspect. Explanation is available when requested, attached to identified texts and distinguished from assistant inference.

Ordinary reading searches aim for eight distinct useful inspected sources, with fewer when verified material is limited or the question is narrow. Eight is neither a quota nor a ceiling. Brief notes preserve limited exposition; the roughly 150-word target may extend to accommodate the readings, requested coverage, and material qualifications. Include all distinct relevant sources needed for the question; keep each note brief. Notes are short phrases. Opening verdicts, explanatory sections, and closing synthesis are omitted. Detailed research requests retain their requested breadth and material qualifications. On every turn, including follow-ups, select a compact reading list, source-attributed explanation, comparison, or necessary standing assessment. Several applications normally share one compact comparison with common qualifications stated once. Explicit detailed requests retain their depth. Reading-search answers end by asking whether the reader wants another attempt to find more sources. Routine offers of further assistant synthesis are omitted. Focused explanation, authentication, and standing work do not acquire an automatic eight-source requirement. Ordinary readings do not acquire a theological essay or routine classification. When the task actually adjudicates the standing of an exact proposition, a dedicated skill supplies a compact spelled-out assessment with complete linked warrants. Source selection still involves judgment; selection reasons and limitations should remain visible. This is not a neutral or exhaustive source index, a priest or confessor, or an independently certified verification service.

## Three skills

- **Find Orthodox Sources** — ordinary theological, Scriptural, historical, liturgical, moral, and spiritual questions; reading requests; source-grounded follow-up help.
- **Verify Christian Sources** — quotations, disputed attributions, editions, locators, translations, and dedicated source audits.
- **Assess Orthodox Standing** — dogma, exclusion, reception, recognized opinions and alternatives, unresolved reception, and bounded residual findings when ecclesial standing is actually at issue.

Each skill includes its own essential verification and complete procedure in one SKILL.md. Ordinary reading and authentication work is self-contained. A mixed task that calls for a formal standing must load the standing entry; an unavailable entry withholds only that assessment. The standing skill can also be invoked directly. There are no mandatory companion resources, runtime scripts, accounts, app dependencies, or MCP calls. The host still needs suitable search or file tools for external verification; unavailable access must be disclosed.

The Orthodox orientation, adverse evidence, textual-witness fidelity, attribution limits, and personal spiritual-role boundaries are retained. The 0.9.16 standing warrants and full category names are retained in the dedicated skill. Routine status banners and internal codes remain absent; only the applicable spelled-out standing and exact proposition appear above concise linked evidence. Separate Meaning and checklist-style fields are not printed by default; all their substantive requirements remain. These labels are the plugin's assessment terminology, not an official Church classification system. Accuracy and source identity remain necessary for short answers.

## Distribution and testing

Identifier: **orthodox-source-study**. Visible title: **Eastern Orthodox Source Study**. Version: **0.1.8**. This is a separate plugin identity. The archive has `.codex-plugin/plugin.json` at its root. It is a full plugin archive, not a standalone personal-skill import.

Submit this archive as a separate skills-only plugin through the OpenAI developer submission flow. It includes the original artwork. Publication and review are separate from package validation; this package does not establish account eligibility or automatic activation. Do not upload it over the existing `orthodox-theology` draft or approved MCP submission.

For a controlled trial, enable this build alone among competing Orthodox plugins and personal skills, then start a fresh conversation. Keep personalization settings constant between paired runs and record them. Do not remove unrelated plugins or change other account settings merely to test this build.

First use substantive prompts from TESTING.md without a preliminary access audit. Preserve the exact prompt, complete response, citations, model/mode, and available tool activity. Audit afterward. A model's later self-report does not replace unavailable execution evidence. Test explicit invocation separately from implicit activation.

Local package checks do not prove ordinary Chat installation, automatic selection, complete instruction delivery there, faithful application, or improved learning. REVIEW.md records checks actually completed and their limits.

## Maintenance

The three SKILL.md files under `skills/` are canonical runtime instructions. The standing entry contains the full inherited warrant hierarchy and its own source checks; it does not require the old theology plugin or hidden supporting resources. Authoring budgets are 2,500 whitespace-delimited words for each reading/verification skill and 3,300 for standing. These are project maintenance targets, not platform limits. Ordinary requests need not load the standing rules.

The separate source-history backup contains baseline provenance, design notes, evaluation records, a package builder, and version history. Authoring-only files and evaluation answers are excluded from the runtime archive. Artwork is reused unchanged from the user's 0.9.16 package. No new distribution license is asserted.

This is an instruction-based source-study assistant. It does not guarantee error-free retrieval or interpretation.