Collect, Store, Analyze, and Act
See every agent session. Then fix what’s broken.
HypAware records every session from Claude Code, Codex, Cursor, and more, and makes all of it searchable. That’s the observability. Then it reads those sessions for you and hands back what’s broken, the evidence, the fix, and whether it stayed fixed. The improvement loop for all your agents.
Collect.
A small daemon on your machine hooks into Claude Code, Codex, Cursor, OpenCode, Pi, OpenClaw, Hermes, etc. and records every session. Nothing else to instrument.
Store.
Sessions land as open table files, indexed and linked into a graph of what touched what: sessions, files, tools, skills. No warehouse, no ingestion pipeline.
Analyze.
SQL and grep narrow millions of sessions to the ones that matter. Then a model reads those and comes back with findings, each with linked evidence, and the root cause fix for each.
Act.
Your agent applies the fix. The next pass checks whether the finding is still happening.
01Observability
Everything you’d want from observability.
Every prompt, tool call, error, and retry, from Claude Code, Codex, Cursor, OpenCode, Pi, OpenClaw, Hermes, and more. Nothing to instrument: the sessions already on your machine are readable on day one. Search them with SQL and grep, from the terminal or from your agent.
02The loop
Then the part observability leaves to you.
- Evidence from every session. The sessions behind each finding, including the jobs nobody was watching. Open any of them and check.
- What’s broken, and how often. What your agents spun on, retried, and threw away, and how many times.
- The fix, not just the symptom. A root-cause recommendation for each finding: a new skill, a missing tool, a permission, an environment issue. Your agent can act on it directly.
- Whether it stayed fixed. Agents change, models change, skills rot. Every pass checks the last one’s fixes, so a fix that quietly stopped working gets caught, not assumed.
Only shows you what happened, and you have to figure out how to fix it.
Finds the root cause, hands your agent the fix, and checks whether it held.
03Your team
One view across the whole team.
With HypAware Cloud, every developer’s sessions, the software factory, and the jobs that run unattended land in one place, with findings across all of them.
Start with the sessions already on your machine.
npm i -g hypaware hyp setup
Install HypAware using npm i -g hypaware, then run hyp setup --guide and follow its instructions.Free and local to start. HypAware Cloud brings in the rest of the team.
Requires Node.js 22.12 or higher, on macOS or Linux.