Coming Soon — Akashic is in active development
Sign up Product Pricing Install Pitch deck AI features 16 phases
Open portal VS Code
v0.3.63 — Knowledge Graph + Codebase Graph · Prefect · Run History · Proprietary Deployments

Code intelligence
in your browser
or VS Code

Use Akashic entirely from the web portal or from the VS Code extension — same Apollo backend, your choice. Batch audit, incremental checks, and interactive agent runs; Prefect-orchestrated phases with run history and manifests. Local .akashic plus durable tenant-scoped cloud storage. Knowledge Graph, tickets, and CI/CD.

Web UI or extension A Colossal Capital Product Analyze → Execute → Deploy Fully managed Apollo backend
Akashic — Analysis Complete
Tools
Mount Directory
Apollo Chat
Semantic Search
Analysis
Full Analysis
Index Codebase
Analyze Code
Analyze Docs
Integrations
Apollo
Slack
GitHub
GitLab
Linear
Cloud
AWS
GCP
Azure
1// Analysis output
2phases: 16 complete
3tickets: 23 generated
4ci: "Tekton (Kaniko, Trivy, Cosign)"
5cd: "Argo CD, Kubes, Bicep"
6doc_coverage: 78%
7security: 0 critical, 3 flagged
8knowledge_graph: "synced + hosted"
9docs_site: "MkDocs built"
10cloud: "AWS connected"
11readiness: "Operational" // green
Written to .akashic/

Apollo Integration
Two tools, one intelligence layer

Akashic is your surface: the web portal (full product in the browser) or the VS Code extension. Apollo is the managed AI backend. Together they bring deep codebase intelligence wherever you work — enterprise-grade security and privacy built in.

Akashic
Web portal + VS Code. Run analysis, view run history, chat, semantic search, CI/CD, ticket management, and cloud deployment from the browser or the editor — live status indicators for every integration.
Web portal VS Code Prefect Chat / RAG CI/CD AWS / GCP / Azure Status Glow
REST API
Bearer Auth
Apollo
The managed AI backend. LLM routing via Together AI, RAG indexing with Weaviate, ticket generation, PR validation, and custom training loops.
FastAPI Together AI Weaviate Terraspace

AI Features
Intelligence built into your sidebar

Beyond analysis phases — Akashic brings a full AI-native toolkit into VS Code, built on Apollo and Together AI.

Q
Apollo Chat + Semantic Search
Ask questions about your codebase in natural language. Apollo answers using RAG-indexed context from your actual code via Weaviate — not generic LLM knowledge. Search by meaning, not just text.
RAGWeaviateCross-file
KG
Knowledge Graph + Codebase Graph
Two graphs, built automatically from any language. The Knowledge Graph maps entities and relationships across your org — rebuilt after every PR merge, viewable in the hosted Graph Explorer. The Codebase Graph is a structural call graph with ripple-risk edges, grounding every ticket with precise dependency context. TypeScript, Go, Python, Rust, Java, C#, and more.
Graph ExplorerCall GraphDrift Detection
T
Together AI — Routing + Fine-Tuning
Apollo's 5D Smart Router picks the best model per task. Together AI for fast inference, Hugging Face TGI on K8s for VPC-isolated inference. Telemetry from every run feeds into LoRA fine-tuning — Apollo gets smarter on your codebase over time.
Smart RouterLoRAVPC TGI
AT
Autonomous Ticket Execution
Apollo decomposes tickets into sub-tasks using full graph context — real file paths, ripple-risk, and dependency edges. You approve, Apollo writes code, runs tests, and opens a draft PR.
Ripple RiskSub-tasksAuto PR
DT
CI/CD + Deployment
Tekton Pipelines run CI in your VPC — Kaniko builds, Trivy scans, Cosign signs, SLSA attestation. CD auto-detects your target: Argo CD, Juju, Kubes, Bicep, CloudFormation, or Terraspace.
TektonArgo CDTerraspace
W
Background Watchers
Watches for code and doc changes in the background. Automatically creates tickets, updates READMEs, and triggers PM reconciliation with Jira and Linear — keeping everything in sync.
File WatcherAuto-docsPM Sync
MR
Multi-Repo Analysis
Mount a parent directory and Akashic auto-detects every nested repo. Cross-repo analysis identifies integration points and builds a consolidated Knowledge Graph via Weaviate.
Auto-detectCross-repoWeaviate
SW
Setup Wizard — Local or Cloud
One command deploys the full stack. Local: Docker Compose, Kind + Docker, Kind + Podman, or Terrakube (Terraform platform). Cloud: Terraspace → AKS / EKS / GKE. Includes Destroy Stack to tear everything down.
Docker ComposeKindGaiaTerraspace
PF
Prefect Orchestration
Background workflows on Prefect — nightly graph rebuilds, 30-minute PM sync, and weekly training jobs. Observable, retriable, and deployable to any work pool.
PrefectScheduled FlowsObservable
RH
Run History
Every analysis, graph build, ticket execution, and deployment is recorded in the sidebar. Filter by type and status, expand for step-by-step breakdown with timestamps.
Run LogsStep BreakdownPersistent
Atlas Integration
Push analysis outputs, tickets, and documentation updates directly into Atlas's multi-entity system — keeping every project in sync across your org.
AtlasMulti-entity PM
Cloud + Integrations
AWS, GCP, Azure, GitHub, GitLab, Bitbucket, Linear, Jira, Slack — all with dedicated sidebar panels, masked keys, live status indicators, and one-click connect/disconnect.
AWSGCPAzureSlack
Q
Apollo Chat — RAG-indexed context
Y
You
Where is authentication handled and what middleware protects the API routes?
Apollo
Authentication is handled in src/middleware/auth.* — a middleware layer that validates API keys on every request. Exempt paths include /health, /docs, and /api/auth/*. Keys are managed via api_key_manager with scoped storage. Works across any language — TypeScript, Go, Python, Rust, Java, and more.
Y
You
Which models does Apollo use for code analysis vs chat?
Apollo
Code analysis phases use DeepSeek-R1 for reasoning and Llama-3.3-70B for generation. Chat uses Mixtral-8x7B for low-latency responses. All routed through Together AI — you can override per-request via the model router.

Analysis Engine
16 AI-native analysis phases

Trigger full or incremental runs from the web portal or VS Code. Run all phases at once or individually, across any language or framework. Every output lands in .akashic/ — versioned, readable, shareable — with run history in the portal when using managed cloud.

01
File Classification
Categorize files, detect project type, map architecture and tech stack.
02
Detect Duplicates
Find duplicate code blocks, redundant files, and copy-paste patterns.
03
Cleanup Analysis
Dead code, unused imports, stale dependencies, safe-to-remove files.
04
Analyze Code
Security vulnerabilities, code quality, architecture, and test coverage.
05
Scaffolding
Missing structure, recommended patterns, and boilerplate gaps.
06
Analyze Docs
Documentation coverage, quality scoring, and inline comment gaps.
07
Cross-Repo Scan
Shared patterns and dependencies across multiple repositories.
08
Docs Consolidation
Merge scattered docs into a structured knowledge base.
09
Script Intelligence
Analyze build scripts, CI configs, Makefiles, and automation.
10
Folder Intelligence
Isolated modules, undocumented folders, and integration gaps.
11
Generate Tickets
Atomic, prioritized dev tickets from all analysis findings.
12
Knowledge Graph
Entity relationships, dependency graph, and architectural map.
13
Deployment Configs
Docker, K8s, Podman, Terraform, and Terraspace generation.
14
Understand Repo
Holistic AI understanding — purpose, patterns, and tech debt.
15
Restructuring Plan
Comprehensive plan synthesized from all preceding phases.
16
Drift Analysis
Semantic drift detection since the last analysis run.

CI/CD Automation
Tekton CI + GitOps CD in your VPC

Tekton Pipelines run CI in a VPC — private connectivity to your services. GitHub, Bitbucket, and GitLab all push to Tekton, keeping CI in sync regardless of Git provider.

V
PR Validation
Every PR reviewed by Apollo against open tickets. Matches, concerns, and recommendations posted as a PR comment.
E
Ticket Execution
Trigger Apollo to implement a ticket autonomously — generates code changes, pushes a branch, opens a draft PR.
K
One-click Credential Push
Auto-register your Apollo URL and API key as secrets in GitHub Actions and Bitbucket Pipelines.
S
Ticket Sync
Push generated tickets from .akashic/tickets/ directly to Linear, Jira, or GitHub Issues.
KG
Post-merge Graph Rebuild
On every PR merge, Apollo rebuilds the Knowledge Graph, regenerates docs, marks cloud tickets Done, and pushes everything back to the repo with [skip ci].
🚀
Auto-Deploy Trigger
After reconciliation passes, Apollo triggers CD automatically — Argo CD, Juju, or Kubes for K8s; Ufo for ECS; Bicep, CloudFormation, or Cloud Deploy for managed PaaS; Blossom for self-hosted; or Terraspace for any cloud.
Prefect-Orchestrated Runs
Every Akashic analysis run, graph rebuild, PM sync, and training job executes as a Prefect flow — scheduled, observable, and retriable. Nightly graph rebuilds, 30-minute PM sync, and weekly fine-tuning run on Prefect work pools with full run history in the portal and the VS Code sidebar.
.akashic/ci/github-actions-apollo.yml
# Tekton CI Pipeline (VPC-isolated)
✓ Kaniko build → image pushed
✓ Trivy scan → 0 critical, 0 high
✓ Cosign sign → image verified
✓ Syft SBOM → attached
✓ SonarQube → 92% coverage, A rating
✓ Tekton Chains → SLSA L3 attested

# Post-merge → Apollo
✓ Rebuild graph + push [skip ci]
✓ Regen docs + mark tickets Done
✓ Argo CD sync → deployed

run_id: a3f8c2d1 complete

Repo-Level AI Agents
Every repo gets its own AI agent

Akashic creates a dedicated AI agent for each repository — trained on the code, its history, and its context. Interface with it through the MCP server like an LLM that truly understands your codebase.

🧠
Trained on Your Repo
Each agent is fine-tuned on the repo's source code, change history, Knowledge Graph, code map, generated tickets, and project plan. It builds a deep, evolving understanding of your codebase — not generic LLM knowledge.
🔧
MCP Tool Interface
Repo agents are exposed as MCP tools via Apollo's MCP server. Any MCP-compatible client — Claude, Cursor, Windsurf, Cline — can call your repo agent to query context, make code changes, generate tickets, or execute tasks.
Autonomous Code & Ticket Changes
Ask the repo agent to implement a feature, fix a bug, or refactor a module — it writes code, creates branches, opens PRs, and manages tickets with full graph context. Like having an LLM that can actually ship code.
🔗
Composable Across Your Org
Repo agents are registered as composable tools — cross-repo queries, multi-repo refactors, and org-wide analysis all work through the same MCP interface. One agent can call another to coordinate changes across repositories.
📈
Continuously Improving
Every analysis run, ticket execution, and PR merge feeds telemetry back into the agent's training loop via Prefect. The agent gets smarter about your repo's patterns, conventions, and architecture over time.
mcp.json — repo agent as an MCP tool
// Ask your repo agent a question
tool: apollo/repo_agent
repo: "my-org/backend-api"
action: "query"
prompt: "How does auth middleware work?"

// Or make changes autonomously
tool: apollo/repo_agent
repo: "my-org/backend-api"
action: "execute"
prompt: "Add rate limiting to /api/v1/*"

Branch created: akashic/rate-limit
Draft PR opened with full context
Ticket updated in Linear
Repo agent training pipeline (Prefect)
Code corpus indexed (2,847 files)
Change history 1,204 commits ingested
Knowledge Graph 4,391 entities, 12,088 edges
Code map call graph + ripple risk
Tickets 89 tickets, 34 executed
Project plan synced from Linear
LoRA fine-tune epoch 3/3 complete
MCP registered apollo/repo/backend-api

How It Works
The full development lifecycle, automated

From analysis to deployment — every phase connected. Click a tab to explore a specific flow, or view the full pipeline end-to-end.

process-flow.mmd — Full Pipeline

Integrations
Connect your entire stack

Connect from the web portal or VS Code. Extension users: credentials live in VS Code's encrypted secret store — never in plain text, never committed to git.

GH
GitHub
Link repos, sync tickets to Issues, auto-register CI secrets, and PR validation workflows.
BB
Bitbucket
Link repos, view PRs and pipelines, push pipeline variables, and CI automation.
GL
GitLab
Link repos, trigger pipelines, sync Merge Requests, push CI/CD variables, and webhook automation.
Li
Linear
Push generated tickets to Linear projects with priority, labels, and cycle assignment.
Ji
Jira
Create issues from Akashic tickets, sync status back to .akashic/ automatically.
Apollo Backend
Self-hosted AI backend — deploy anywhere with Terraspace. All 16 analysis phases and chat run through Apollo.
Atlas
Push tickets, docs, and analysis to Atlas’s multi-entity management system for automated PM.
#
Slack
Real-time notifications for analysis, PRs, tickets, and training. Link channels per-directory with Block Kit messages.
AWS
Store Access Key + Secret securely. Set region, bootstrap IAM roles, and deploy to EKS or App Runner.
GCP
Service Account JSON or file upload. Set project ID, region, and deploy to GKE or Cloud Run.
Azure
Service Principal credentials. Set subscription, tenant, client ID, and deploy to AKS or Container Apps.

Get Started
Up and running in 4 steps

Sign up for Colossal Capital / Apollo, then use either the Akashic web portal (no install) or the VS Code extension — connect repos, run analysis, generate tickets, and execute.

1
Sign up & choose your surface
Create your account at the Colossal Capital keys portal. Then use the web portal or the VS Code extension — same API key, same backend.
2
Self-host or managed (optional)
On VS Code, run Setup Stack for local runtimes — Docker Compose, Kind + Docker/Podman, or Terrakube — or deploy to AKS, EKS, GKE, Cloud Run, Azure Container Apps, Blossom. Portal-first users typically stay on Colossal Capital managed Apollo.
Cmd+Shift+P → Akashic: Setup Stack
3
Connect & analyze
Connect GitHub, Bitbucket, or GitLab. Add Linear or Jira. Link Slack and your cloud (AWS, GCP, Azure). Mount or select repos — run full batch, incremental, or interactive agent flows; Prefect orchestrates phases with run history in the portal.
Portal UI · or VS Code sidebar
4
Execute & Auto-Rebuild
Approve generated tickets, Apollo executes them — code gen, tests, PRs. On every PR merge, the Knowledge Graph and Docs Site auto-rebuild via Tekton.
PR Merged → Graph + Docs auto-update
Pricing
Akashic is free with Apollo

Akashic is included at no extra cost with any Apollo subscription. Your Colossal Capital tier controls API limits, training budgets, and feature gates across Akashic and every other CC product.

Sign up — CC keys portalView tier details

CC tiers start at $0/mo. Akashic is free with Apollo ($4/mo).


Colossal Capital
Start understanding your codebase today

16 analysis phases, Knowledge Graph, autonomous execution, and CI/CD — from the web portal or VS Code. No extension required if you prefer the browser.

Open web portal VS Code setup CC pricing