℗ 2026 Colossal Capital · managed SaaS
Connect any Git repository — legacy monolith, greenfield service, Cursor project, or a Claude Fable you vibe-coded in one session. Each repo is a Collection. Group related repos into a Library. Span entities in a Workspace. Then mix and publish from the studio with your team and identity-provisioned AI agents.
Fables aren't every repo. They're the new Claude output format — sprawling one-shots that work once but need a path to iteration. Akashic Records is built for that moment and for every other codebase you already have.
one pressing · record · liner · coffee-table book · five source layers
Claude Fables · the timing
Anthropic just shipped Fable — and teams are generating complete applications in a single session. Those repos are real code with real architecture buried inside a vibe-coded spike. They aren't the only thing Akashic handles, but they're the urgent case: a massive one-shot that has to become iterable, team-owned, and deployable.
A Claude Fable — or any similar one-shot AI build — is a repo born from a single creative session. It runs, it impresses, and then it sits there because nobody knows what to do with it next.
Any repository you connect — Fable or not — mastered as one pressing: a record (code), a liner (voice + provenance), and a coffee-table book (tickets + docs).
What a Collection holds
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 (voice + skillfiles), and the coffee-table book your team actually reads (tickets, docs graph, reconciliation). All three draw from the same five source layers — repo, voice, tickets, documentation graph, and codemap.
RAG index, analysis passes, codemap, deployments — what producers and CI run against.
Session provenance — voice commands, prompts, derived SKILL.md profiles. Fables arrive heavy here.
emit_skillfiles — conventions agents inheritTicket matrix, doc navigation, graph explorer, reconciliation gate — the PM artifact product and eng share.
This maps to how Akashic is built. Nineteen Prefect mastering phases populate each artifact — not decorative labels. Mature repos fill the book first; Claude Fables flood the liner until tickets catch up. Same pressing, different balance.
Product structure
Three terms in the product. A Collection is one pressing — record, liner, book. Library and Workspace scale how many pressings you work with at once.
One repository — mastered into record, liner, and coffee-table book from repo, voice, tickets, documentation graph, and codemap.
Multiple Collections under the same entity — API, web, contracts, infra — shared run history and one reconciliation gate.
Collections across different entities in one collaborative environment — your org, a client, a portfolio company.
The studio is the collaborative portal (web or VS Code) where your team works inside a Workspace. Product name stays Workspace; the studio is the room you work in.
Libraries
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.
library · payments · 4 collections
Relationships span every Collection in the Library — see ripple effects before anyone merges.
Tickets dedupe against Linear, Jira, and GitHub across the whole Library. One gate.
Agents per Collection with shared Library context — the whole album, not one track.
The studio · your Workspace
The studio is Akashic's collaborative cloud portal and VS Code extension. Engineers run analyses and review agent PRs; product approves at the reconciliation gate — whether the Collection started as a Fable or shipped years ago.
Mastering passes, Prefect run history, RAG chat, ticket matrix, agent approvals, deployments — no install.
Open your Workspace →Connect Collections, run analysis, semantic search — step into the full studio when you need the whole Workspace.
VS Code Marketplace →Answers from indexed Collections via Weaviate — file paths, not model memory.
SKILL.md profiles from each Collection's conventions, stored in your Workspace.
Tekton pipelines: builds, scans, signing, SBOMs — routed to your targets.
Approvals for syncs, mutations, and deployments — policies you control.
Mastering pass
Run full_analysis from the studio or VS Code. Prefect on Apollo writes into your pressing: code phases on the record, provenance on the liner, PM and docs on the book.
Index, classify, analyze, deploy — the executable substrate.
Voice sessions + emit_skillfiles — how it was made.
Docs, PM context, tickets, archive — what teams reconcile.
Producers
Dedicated AI producers on each Collection — MCP tools through Apollo, Auth0 machine identities, scoped OAuth, Cloudflare Signed Agents audit trails.
Tickets → code → tests → draft PRs on akashic/*.
Coverage checks, regressions, gap tickets when quality drops.
Rebuilds graph and docs archive after every merge.
Get started
Membership
Akashic itself is free — you pay for the Apollo tenant that runs your studio, and for the LLM tokens you actually consume. No hidden markup when you bring your own keys.
The Akashic product — Collections, Libraries, Workspaces, nineteen-phase mastering, ticket reconciliation, and the studio portal — ships at $0. Apollo is the managed backbone at $5/mo per tenant.
RAG chat, analysis passes, and agent work consume model tokens. On Apollo's shared keys, upstream inference cost is passed through to your account — you see what the provider charged, not an opaque bundle.
Connect your own API keys in the studio. Inference routes through your account; Akashic never resells your quota.
Colossal Capital tier sets limits across every product — API calls, repos, training jobs, team size. Tier details · sign up