Generate security guidelines for AI coding agents.
go install github.com/eliate-security/eliate/cmd/eliate@latestAnalyze a codebase and generate AGENTS.md with security guidelines:
export ANTHROPIC_API_KEY=your-key
eliate guidelines /path/to/projectReview code against the guidelines in AGENTS.md and write findings to REMEDIATION.md:
eliate review /path/to/projectBoth commands support:
--timeout <seconds>- Analysis timeout (default: 300)--quiet, -q- Suppress output
The agent operates with restricted permissions:
Blocked file reads:
.env,.env.**.pem,*.key,*.p12,*.pfxid_rsa,id_ed25519credentials.*,*.netrc,.npmrc,.pypirc
Allowed bash commands (read-only):
ls,find,tree,cat,head,tail,wc,file,stat,du,df
Write permissions:
guidelinescommand: onlyAGENTS.mdreviewcommand: onlyREMEDIATION.md