These mission bundles are autoresearch-ready pilots for this repo snapshot.

Each mission directory contains:

• mission.md — objective, scope, and expected deliverable
• sandbox.md — evaluator contract plus safety/operating rules

Current pilots / examples:

• cli-discoverability-pilot/
• security-path-traversal-pilot/
• in-action-cat-shellout-demo/ — a small self-hosted OMX optimization demo that removes the autoresearch loop's manifest cat shell-out and proves the fix with a focused evaluator

You can run the evaluators directly today:

node scripts/eval-cli-discoverability.js
node scripts/eval-security-path-traversal.js
node scripts/eval-in-action-cat-shellout-demo.js

To see a real end-to-end run, launch:

omx autoresearch missions/in-action-cat-shellout-demo

Then inspect .omx/logs/autoresearch/<run-id>/manifest.json, candidate.json, and iteration-ledger.json to see the supervisor's keep/discard/stop decisions.

These bundles are designed to be first-class omx autoresearch missions rather than generic prose examples.