GitHub
Commits, issues, pull requests, releases, wiki and repository documents such as READMEs and changelogs.
Configure during integration
Choose repositories, grant access and set how much history to import.
Connect your source systems and AI tools.
Choose which records to import and which tools your team will use to query them. We agree the connection and configuration work during scoping.
01 · Bring the record together
Source connections.
The platform includes the connector templates below. During integration, you select the source history and configure credentials, field mappings and access rules. Each connection is tested against your sources.
CONNECTOR TEMPLATE · CONFIGURATION REQUIRED
Azure DevOps
Git commits, work items, pull requests and wiki records.
Configure during integration
Configure project access and source identity, then choose which records to extract.
Documents & file shares
Import documents from selected file shares.
Configure during integration
Agree the file formats, source metadata and access classification for the selected documents.
Ticket attachments
Files attached to tickets become documents of their own, linked to the ticket and readable only within its access scope.
Configure during integration
Parsing uses a configured document or image model. Downloads come only from hosts you allow; oversize files are skipped and reported.
Recordings
Recording segments and their source references.
Configure during integration
Agree how recordings are obtained and transcribed, who can access them and what consent is needed.
Deployment configuration snapshots
Configuration captures from your environments, compared over time. Secrets are dropped before anything is indexed.
Configure during integration
Needs a capture process and the deployment-configuration component enabled. Keys, modules and defaults are indexed, never values.
Case-management records
Import cases from the selected case-management system.
Configure during integration
The source API, field mapping and authentication need integration-specific configuration.
Legacy ticket exports
Import ticket history from a legacy archive.
Configure during integration
Check the export format, record provenance and identity mapping against the archive.
Each connector needs testing against the source system and version you use; listing it here does not imply certification.
02 · Work through a standard
MCP server, skills and agent processes.
Three layers. Use the engine through MCP alone, add skills for recurring tasks, then supervised processes where coordination helps.
MCP 2.0
The engine implements MCP 2.0, the stateless generation of the Model Context Protocol, over stdio and streamable HTTP. Hosts on the earlier protocol can still connect. It exposes 42 tools, and the caller’s identity decides which sources each call may read.
THE BASE LAYER108 skills in 13 packs
Each skill is a procedure for one task: its inputs, tool sequence, checks and stop conditions. Investigating a behavior, preparing a change review and guiding an onboarding task are three of them. Together the skills use all 42 tools.
OPTIONAL23 roles, 62 processes
Supervised processes coordinate skills across named roles and come packaged as 19 installable solutions. Every agent step runs one named skill.
OPTIONALSkills and processes are packaged for Codex, Claude Code, GitHub Copilot CLI, Cursor, Gemini CLI, OpenClaw, DeepSeek Harness, Pi, Grok Build and OpenCode, and install as plain skill directories in other Agent Skills hosts. A package is an installation path, not an endorsement or certification by the client vendor.
Counts are from the product catalogues, October 2026. A deployment can expose fewer tools, depending on its configuration and permissions.
Which sources would you connect first?
Tell us which systems hold your code and its history. We can outline the work needed to connect them.