Korg Ledger
Maintain a tamper-evident JSONL session ledger for agent work, with optional HMAC proof, audit, and import of third-party logs.
Overview
korg-ledger is an MCP server for the Operate phase that provides tamper-evident agent session ledgers with audit and import tooling.
What is this MCP server?
- PyPI korgex MCP subcommand mcp-server v0.6.3 over stdio
- JSONL journal defaulting to .korg/journal.jsonl via KORG_JOURNAL_PATH
- Optional KORG_LEDGER_HMAC_KEY for tamper-proof HMAC chaining beyond tamper-evident hashes
- Audit and import paths for cognition logs from other agent vendors
- Positions verifiable cognition: prove ledger integrity to stakeholders or future you
- Server version 0.6.3 via PyPI package korgex
- Required runtime positional argument: mcp-server
- 2 documented environment variables: KORG_JOURNAL_PATH and KORG_LEDGER_HMAC_KEY
Community signal: 1 GitHub stars.
What problem does it solve?
Agent transcripts vanish or get edited and you cannot prove what tools ran when a deployment or billing dispute surfaces.
Who is it for?
Solo builders running multiple AI coding agents who need a local, provable audit trail of cognition and tool use.
Skip if: Teams wanting only cloud SaaS observability with no local journal file or no interest in cryptographic integrity options.
What do I get? / Deliverables
You get an append-only, verifiable journal with optional HMAC proof and MCP audit/import so session history stands up to review.
- Local JSONL session ledger with integrity verification
- MCP-driven audit and import of agent cognition logs
- Documentable proof posture for disputes or compliance reviews
Recommended MCP Servers
Journey fit
Long-running agent sessions need operational visibility first; Operate monitoring is where verifiable logs earn a permanent shelf. Monitoring covers append-only journals, integrity checks, and cross-session audit—not feature coding.
How it compares
Tamper-evident agent ledger MCP, not a generic APM dashboard or skills.sh skill package.
Common Questions / FAQ
Who is korg-ledger for?
Builders and small teams who treat agent sessions as production events and need auditable, importable JSONL ledgers.
When should I use korg-ledger?
Use it when operating or shipping agent-heavy workflows and you must verify logs were not tampered with or merge logs from other agents.
How do I add korg-ledger to my agent?
Install korgex from PyPI, configure MCP stdio with positional argument mcp-server, set KORG_JOURNAL_PATH and optionally KORG_LEDGER_HMAC_KEY, then register io.github.New1Direction/korg-ledger.