Skip to content

Log Decisions & Discoveries

Log an entry

At any point during a session, run:

/log-entry

ml-journal infers the entry type from conversation context and presents a draft for confirmation.

Entry types to know

The most common types you'll log during an investigation:

Type When to log Confirm required?
decision User confirms a direction ("let's do X") Yes
discovery Unexpected finding changes understanding Yes
issue Bug or inconsistency identified No
resolution Fix verified working No
lesson Root cause understood No
experiment Verdict is clear (confirmed/refuted/inconclusive) Yes

For the complete list, see Entry Types & Schema.

Chain patterns

Entries chain naturally during a session. ml-journal auto-proposes each step:

Bug fix chain:

  1. Bug identified → auto-propose issue
  2. Fix verified → auto-propose resolution
  3. Root cause understood → auto-propose lesson

Investigation chain:

  1. Hypothesis formed → manual hypothesis
  2. Experiment run → auto-propose experiment
  3. Unexpected finding → auto-propose discovery
  4. Direction confirmed → auto-propose decision

Tips

  • Log as events happen, not in a batch at the end
  • Accept or decline auto-proposals — ml-journal won't re-propose if you decline
  • Use relative dates in conversation; ml-journal converts to absolute timestamps
  • Link related entries — resolutions link to issues, experiments link to hypotheses

You have now...

Learned to log decisions, discoveries, and issues with /log-entry and let chain patterns guide the audit trail through natural investigation stages.