praxis-memory@0.6.0

Give Claude Code a memory. Never re-explain your project again. Local-first — no server, no account, nothing leaves your machine.

AI Security Review

scanned 2h ago · by lpm-firewall-ai

LPM treats this as warn-only first-party agent extension lifecycle risk. The package establishes a first-party Claude Code extension: project hooks, project commands, and optional user-scope slash commands. Its hooks read local Claude transcripts and write derived PRAXIS memory/state files; opt-in telemetry sends only aggregate counters.

Static reason

No blocking static signals were detected.; previous stored version diff introduced dangerous source

Trigger

User runs `npx praxis-memory` or `praxis init`; subsequently configured Claude hook events invoke `praxis capture` and `praxis tray --ensure`.

Impact

Persistent AI-agent integration can read session transcript files and write local project/user Claude configuration, but no unconsented install-time execution or confirmed content exfiltration is present.

Mechanism

Explicit CLI setup of Claude hooks and local transcript-derived memory persistence.

Rationale

Not malicious by source evidence: configuration and execution occur after an explicit CLI invocation, not during npm installation, and no concrete exfiltration or destructive chain was found. Per policy, the first-party agent extension lifecycle and global Claude command setup warrant a warn verdict.

Evidence

package.jsonsrc/cli.jssrc/commands/init.jssrc/lib/settings.jssrc/commands/capture.jssrc/lib/telemetry.jssrc/commands/tray.jssrc/tray/tray.ps1.claude/settings.json.claude/commands~/.claude/commandsCLAUDE.md

praxis-memory@0.6.0

AI Security Review

Decision evidence

Source & flagged code

Source & flagged code

Findings

Source & flagged code