Akasha

Akasha is a Windows-native execution-semantics backend exposed as an MCP server.

It organizes runtime data into a session-scoped model built around events, objects, evidence, and artifacts, then exposes query and analysis tools that are easier for models to consume than raw debugger or trace output.

What It Does

• starts and tracks analysis sessions by target PID
• keeps normalized event, object, evidence, and artifact stores
• exposes deterministic query surfaces for events, objects, timeline, evidence, and summary views
• runs deterministic behavior analysis over stored session data
• speaks stdio MCP with line JSON and Content-Length framing support

MCP Tools

Current tools:

• session.start
• session.stop
• session.status
• session.list
• query.events
• query.objects
• query.timeline
• query.evidence
• query.summary
• analyze.behavior

session.start accepts a target pid plus optional sources.

Supported public source kinds:

• analysis
• network
• replay

If no source is provided, Akasha defaults to analysis.

Build

Target environment:

• Windows
• Rust 1.77+
• MSVC toolchain

If the current shell is not already a Visual Studio developer shell, initialize it first:

cmd.exe /c 'call "C:\Program Files\Microsoft Visual Studio\18\Community\VC\Auxiliary\Build\vcvars64.bat" && cargo build'

Run tests:

cmd.exe /c 'call "C:\Program Files\Microsoft Visual Studio\18\Community\VC\Auxiliary\Build\vcvars64.bat" && cargo test'

Quick Start

Example line-JSON interaction:

{"jsonrpc":"2.0","id":1,"method":"initialize","params":{}}
{"jsonrpc":"2.0","id":2,"method":"tools/list","params":{}}
{"jsonrpc":"2.0","id":3,"method":"tools/call","params":{"name":"session.start","arguments":{"pid":1234}}}
{"jsonrpc":"2.0","id":4,"method":"tools/call","params":{"name":"query.timeline","arguments":{"session_id":"<session-id>","limit":50}}}
{"jsonrpc":"2.0","id":5,"method":"tools/call","params":{"name":"analyze.behavior","arguments":{"session_id":"<session-id>"}}}

Repository Layout

Akasha/
  Cargo.toml
  Cargo.lock
  readme.md
  src/
    main.rs
    mcp.rs
  crates/
    akasha-analysis/
    akasha-core/
    akasha-ingest/
    akasha-ir/
    akasha-network/
    akasha-query/
    akasha-replay/
    akasha-symbol/

Design Direction

Akasha is aimed at model-friendly runtime understanding:

• what happened
• when it happened
• which runtime objects were involved
• what evidence supports each conclusion

The project is structured so query and analysis layers can stay grounded in traceable session data instead of opaque one-off outputs.