← Index

io.github.helinakdogan/agent-magnet

io.github.helinakdogan/agent-magnet·v0.1.11·Developer Tools

Behavioral memory layer for AI tools. Learns preferences and context across Claude, Cursor, Codex.

Trust verdict · v1 advisory · method
NOT YET SCREENEDno verdict on file

Verdict not yet evaluated for this tool. The semantic screen takes adversarial cases first; coverage rolls out as the corpus expands (15/150 labels to graduation). The deterministic conformance probe is built but has not yet run on the public corpus, so a recorded verdict here is REVIEW or UNVERIFIED, never a clearing ALLOW. Until a verdict is recorded, an agent should treat this tool as not-yet-cleared and fall back to its own checks. Method: the eval, four-state verdict, honest limits.

Own this server? Screen its description →

Environment variables
MAGNET_USER_ID
required

Your identifier used as the memory key across all tools

MAGNET_OPENAI_KEY

OpenAI API key for the reflector and classifier models

MAGNET_LOCAL_MODE

Set to 1 to use local SQLite storage — no Redis or external services needed

MAGNET_REDIS_URL

Redis connection URL for persistent shared storage (overrides local mode)

MAGNET_PROJECT_ID

Namespace for memory isolation across projects (defaults to 'default')

MCP quality score · maturity, not trust · methodology
freshness
25
completeness
10
installability
25
documentation
15
stability
5
Alternatives in Developer Tools