Tools
Production-ready infrastructure for managing AI agent behavior
PromptOps
stable
v0.6.0
Git blame for prompts: incident archaeology, git-native versioning, and a no-SaaS production runtime
- Incident archaeology: blame --at answers what was live when it broke
- Production runtime with no .git/, no git binary, no SaaS
- Semver impact graded from the variable signature, never prose
- One-command health check with promptops doctor
- Version and deploy history in one timeline
- Structured PROMPTOPS_E0XX errors with hints and doc links
PyPI
GitHub
Documentation →
ReleaseOps
early-access
v0.1.0
Release engineering infrastructure for AI agent behavior
- Bundle versioning with SHA-256 content addressing
- Gated promotion through dev → staging → prod
- Automated evaluation with pluggable judges
- OpenTelemetry integration for production observability
PyPI
GitHub
Documentation →
How they work together
# 1. PromptOps versions your prompts
promptops create prompt support-system
git commit # → tags prompt-welcome-v1.0.0 (after hooks install)
# 2. ReleaseOps bundles and ships them
releaseops bundle create support-agent \
--artifact system=support-system:v1.0.0 \
--model claude-sonnet-4-5 --provider anthropic
# 3. Promote through environments
releaseops promote promote support-agent 1.0.0 dev
releaseops promote promote support-agent 1.0.0 staging
releaseops promote promote support-agent 1.0.0 prod
# 4. Trace behavior back to prompts
releaseops attribution explain support-agent 1.0.0 \
--action "approved refund for $120"Get started
Set up both tools in under 5 minutes.