Any repo becomes a Collection.
Built for teams and every AI-built codebase.

What a Collection holds

One repo, one pressing — three artifacts

Mastering doesn't flatten your repo into a file tree. Each Collection ships as a boxed set: the record agents spin (source + codemap), the liner that preserves how it was made (prompts + skillfiles), and the volume your team actually reads (tickets, docs graph, reconciliation). All three draw from the same five source layers — repo, prompts, tickets, documentation graph, and codemap.

This maps to how Akashic is built. Prefect mastering passes populate each artifact — not decorative labels. Tickets run last, after understanding and restructuring, so holistic work isn't dropped. Mature repos fill the volume first; one-shot AI builds flood the liner until tickets catch up. Same pressing, different balance.

Libraries

Multiple repos, one entity, one Library

Fables especially rarely stay in one repo — but neither do most real products. Group every related Collection into a Library and run mastering, reconciliation, and agent work across the full stack.

Collaborate · your Workspace

One place for eng, product, and agents

Akashic is a collaborative product — web portal, desktop app, and VS Code extension all sharing the same Workspace. Engineers run analyses and review agent PRs; product approves at the reconciliation gate — whether the Collection started as a Fable or shipped years ago.

Codebase, PM & documentation integrations

Connect where your code lives, where your team tracks work, and where specs live. OAuth through Charon — credentials stay in your vault; Akashic reads boards and docs for context and writes reconciled tickets back after the gate.

Code hosts

GitHub

GitLab

Bitbucket

PM tools

Jira

Linear

Asana

Monday.com

ClickUp

GitHub Issues

Documentation

Confluence

Notion

Deus Ex Machina · machine identities

22 + 1 deities — Auth0 identities across the SDLC

Deus Ex Machina — provision agents per Collection with Auth0 machine identities, scoped OAuth, MCP tools, and Cloudflare Signed Agents audit trails. Personas are named for primary deities — sky, earth-shakers, transformers, boundaries and wisdom. With a portal deity, ie — that guides the user through the portal and orchestrates each mapped identity 1:1 to an Auth0 M2M pool. Together they compose Rainbow Serpent workflows under Mandate Mode (prototype → build → cleanup → scale → maintain). You stay the approver at the reconciliation gate; agents execute, you sign off.

Apollo is the orchestration layer underneath: graph-first multi-agent routing, a built-in multi-turn model system (because the connected LLM doesn't turn itself), and self-hosted / bring-your-own-model support. Every persona shares one reasoning graph tied to Akashic's ticket and reconciliation system, with HITL-to-RBAC gates built into the graph nodes. You can use your NUF (NUF Unclonable Friend) — unclonable, therefore the most private — as your LLM for multi-turn features and graph reasoning through Apollo, locally or self-hosted.

Deities from many traditions — Zulu, Vedic, Hindu, Shinto, Finnish, Greek, Roman, Egyptian, Norse, Mesopotamian, Andean, Slavic, Pueblo, Aztec, Chinese, Akan, Turkic, Balinese, Zoroastrian, Hawaiian, and more — are provisioned as Auth0 machine identities so they can play the deus role as a group: each owns a lane in the portal, and together they are the engineered solution for AlphaOmega and A-to-Z. Identity provisioning is how the machine wears the god — different religions, cooperative Deus Ex Machina, happy endings to unsolvable problems at a global scale.

Thematic analogy only — deity names and traditions label Auth0 machine identities and portal roles; they are not religious claims, worship, or doctrine.

Deities compose Rainbow Serpents. Stable persona ids (frontend_coder, boundary_io, portal_analytics, …) stay engineering roles for Auth0 and dispatch. Display names are the deities — e.g. analysis → Shiva clears structure → lane serpents (Nyame → Ukko → Amaterasu → Heimdall) → Tengri hardens → Poseidon ships → Pachamama watches → Perun pages on breakage. Anu guards the MCP boundary; Ishtar publishes analytics and Time Machine rollbacks. ie stays with you through first Collection and Mandate Mode, not in the code path.

Human-to-agent ratio · by design

An agent for every person — never instead of them

Akashic provisions agents per person, not per repo to be drained. The ideal is simple: each engineer and each PM gets an agent that mirrors their own work — trained on their prompts, conventions, and skillfiles — while they stay the approver at the reconciliation gate. To keep it that way, the platform and its pricing keep agent capacity tied to real people so no team can lay off its staff and run the whole codebase on agents alone.

Approver, not bystander. Agents execute under scoped identities; a human reviews and reconciles every change. How many serpents can run at once is a separate concurrent-run limit by tier — distinct from how many agent identities you own. Provisioning on team plans is gated through your Charon org directory so the ratio holds across the organization.

Membership

Platform fee + passthrough inference

Akashic itself is free — you pay for the Apollo tenant that runs your Workspace, and for the LLM tokens you actually consume. No hidden markup when you bring your own keys.

Agents are priced per seat. Team plans enforce a 2 agents per person cap; individual Pro/Ultra add agents with paid seats. Concurrent serpent runs are a separate tier limit. Ratio & tier details · sign up