Every AI code action, linked into one tamper-evident chain of evidence.
Pillar 01 · Audit & Accountability
Pramana attributes every code action to the specific AI coding agent, Claude Code, Cursor, Copilot, in a tamper-evident audit trail, then turns it into audit-ready compliance evidence.
Sanskrit: “proof / valid evidence”
Every AI code action, linked into one tamper-evident chain of evidence.
The problem
When an AI agent commits or pushes code, most teams have no reliable record of which agent did it, what it changed, whether it touched sensitive systems, or whether a human reviewed it. So when an auditor asks “how do you govern AI-generated code?”, there is no answer.
A lightweight CLI with git hooks and repository webhooks detects each action and the agent behind it.
Every event is written to a tamper-evident, hash-chained audit trail, attributed to the specific AI agent.
A risk engine flags high-signal events (direct-push-to-main, large diffs, sensitive paths) into a human reviewer queue.
Export SOC 2 / ISO 27001 / ISO 42001-styled compliance evidence reports on demand.
Screens shown with illustrative sample data.
Knows whether Claude Code, Cursor, or Copilot made each change. The differentiator nobody else centers on.
A hash-chained event log, cryptographically verifiable, so the record can’t be quietly altered.
High-signal rules with severity, plus an approve / dismiss queue with role-based access.
Natural-language Q&A over your audit data, “what did Claude Code do on main this week?”
One-click SOC 2 / ISO 27001 / ISO 42001 evidence packs, mapped and export-ready.