Ellora

Capture a context document from this team's content session. It is always saved disabled; an admin must activate it before it can affect rewrite prompts.

ellora_context_capture

Confirm one voice profile is active and usable by voice_profile_id, returning its name. Only active profiles are returned — the same requirement ellora_session_start enforces.

ellora_profiles_fetch_by_id

Confirm one writing brief is active and usable by writing_brief_id, returning its name. Only active briefs are returned.

ellora_briefs_fetch_by_id

Create a saved voice profile for this team from writing samples, and make it active. Use this when the team has no active voice profile yet — check ellora_profiles_fetch_active first. Requires a team admin. Returns the new voice_profile_id, usable with ellora_session_start.

ellora_profile_create

Fetch the full body of one active context document for your team. Read-only — the fetch bumps only an internal retrieval counter, never the document content, kind, source, or status.

ellora_context_fetch_by_id

Return recommendation-only positioning and messaging guidance from this team's active positioning profile, approved context, and learned voice preferences. Read-only — selecting context bumps only an internal retrieval counter, never any governed field.

ellora_positioning_recommendations

List active context documents for your team. Returns metadata only; use ellora_context_fetch_by_id for the full document.

ellora_context_fetch_active

List every active voice profile for your team, with their voice_profile_id — the IDs usable with ellora_session_start.

ellora_profiles_fetch_active

List every active writing brief for your team, with their writing_brief_id — the IDs usable with ellora_session_start — plus each row's nested writing_brief_json, which can be passed back to ellora_session_start verbatim.

ellora_briefs_fetch_active

Analyze this content against the voice profile/writing brief fixed by ellora_session_start. Returns findings plus a convergence hint — follow next_action. When mandatory findings are open, the response includes rewrite_guidance: rewrite the content yourself and call ellora_review again before showing anything to the user. Once a call passes and it is not the session's first, this also automatically verifies the result against the session's original content for regressions and, on success against a saved voice profile, teaches the profile from it. Stop and deliver once you see a verification block with passed: true, or once next_action says diminishing_returns.

ellora_review