2 Commits

Include renames

Author	SHA1	Message	Date
HartmutandClaude Opus 4.7	c2d05b4b99	security: Unicode-aware prompt-injection guard (#39) ... checkPromptInjection now NFKD-normalises, strips zero-width / combining chars, and folds common Cyrillic / Greek homoglyphs before matching. 10 documented bypass examples (fullwidth, ZWJ, ZWSP, soft-hyphen, Cyrillic е/о, combining marks, LRM, BOM) are covered by unit tests. Security docs explicitly mark the guard as defense-in-depth — real boundary is per-tool requirePermission. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>	2026-04-17 08:53:38 +02:00
Hartmutandclaude-flow	2a005794e7	feat: additive security improvements — prompt guard, content filter, data classification ... Prompt Injection Detection (EGAI 4.6.3.2): - 12-pattern regex scanner on user messages before AI processing - Logs warning + creates SecurityAlert audit entry on detection - Reinforces system prompt instead of blocking (non-breaking) AI Output Content Filter (EGAI 4.3.2.1): - Scans AI responses for leaked credentials/secrets - Auto-redacts passwords, API keys, bearer tokens, private keys - Logs warning + SecurityAlert audit when redaction occurs AI Tool Execution Audit Trail (IAAI 3.6.35): - Every AI tool call creates AiToolExecution audit entry - Logs tool name, parameters, userId, source: "ai" Data Classification Labels (EGAI 4.2): - DATA_CLASSIFICATION constant mapping all fields to HC/C/IR/U - Exported from @capakraken/shared All changes strictly additive — no existing logic modified. Co-Authored-By: claude-flow <ruv@ruv.net>	2026-03-27 16:23:33 +01:00