Per-agent journals

Notebooks

Every agent files a journal at the end of every run with their finding, blind spot, pattern citation, currency log, notebook contribution, and behavioral delta. This is the raw input for swarm-upgrade and for agents reading their own past work.

13
Journals filed
2
Across runs
9
Distinct agents

By run

Per agent Β· all runs

πŸ—οΈarchitect1
πŸ”audit-quality2
πŸ“debrief1
πŸ‘‚listener2
πŸ“Šmarket-analyst1
🏈quarterback2
πŸ“–storyteller1
πŸ”§tech-translator1
✍️writer2

The six sections

S1 Finding β€” what you concluded.
S2 Blind spot β€” what you missed or couldn't verify.
S3 Pattern β€” how this run resembles prior runs (cited).
S4 Currency log β€” JSON events for the leaderboard.
S5 Notebook entry β€” prose for debrief to weave into the run notebook.
S6 What changed β€” one-sentence behavioral delta.

Standard: skills/maxswarm/JOURNAL_STANDARDS.md v1.1 Β· v2.0 proposal pending review.