Skip to content

Learn

Collaborator errors → distilled into guide updates. It's a step in our roadmap for improvement


Process

as we roll along, we hit a lot of bumps. i've noticed that i get fed up and stop dead. clean house. takes time. need a better triage system. Let's start with:

  • list mistakes as they happen (oldest last)
    • hyphen-N date title
  • distill: identify pattern, write rule, add to guide
  • research: better tools, clever ideas
  • track for escalating need:
    • fed up
    • stop dead
    • clean house

To distill an entry:

  1. Identify the pattern (what went wrong, repeatedly?)
  2. Write a rule (imperative, actionable)
  3. Add rule to the appropriate guide file
  4. Remove the raw entry from this file

Raw Log

Empty. Everything here has been distilled into a rule; the table below says where each went. What is di's own — its geometry and its pipeline — is in di's learn.


Distilled

PatternRule added to
Stale readsalways.md — "Re-read before editing"
Wrong year/path assumptionsalways.md — "Verify Before Writing"
Tool failure deflectionworkarounds.md — "Tool Failure Recovery"
Incomplete renameworkflow.md — "Rename with mv, then search"
Project-specific swapworkflow.md — "Remove, don't swap"
Worktree pathsalways.md — already covered in "All file paths"
Drifting from requirementsalways.md — "Requirements Echo"
Ignoring shorthandalways.md — "Shorthand First"
Revisit as info not actionshorthand.md — revisit now includes "ask Work on?"
Incomplete journalshorthand.md — journal now says "Execute ALL parts"
Contradicted selfalways.md — "Before saying No, verify"
Said 'stand' for what is still therebanned words.md — remain, unchanged; never stand
Misread exact namealways.md — "Exact names matter"
Checked off without testingalways.md — "Code written ≠ feature complete"
Asked which task firstalways.md — "Task lists are priority-ordered"
Used Bash ls instead of Glob/Readalways.md — "Use Glob/Read, not Bash"
Speculated instead of observingpitfalls.md #12 — "Observe before speculating"
Abbreviated code namespitfalls.md #13 — "No abbreviations in code"
Misread "here" as file referencepitfalls.md #14 — "'Here' means chat output"
Wrote to Claude memory dirCLAUDE.md — already added
Worktree paths (repeated)pitfalls.md #7 — already covered
npm instead of yarn (repeated)always.md — already covered
Skipped always.md reads (repeated)CLAUDE.md — already covered
Miscounted entries (stale read)pitfalls.md #1 — already covered
Ignored shorthand command (repeated)pitfalls.md #8 — already covered
Logged without checking existing rulespitfalls.md #15 — already covered
Treated question as instructionpitfalls.md #16 — "Questions aren't instructions"
Promised beyond this chatpitfalls.md #17 — "Scope promises to this chat"
Trimmed rejected idea instead of restartingpitfalls.md #18 — "Rejection means start over"
Treated analysis shorthand as actionpitfalls.md #19 — "Analysis shorthands are not action requests"
Guessing said as factalways.md #3 — "Guess"
Reading the log myselfresponse.md #7 — "Read the log yourself"
Formatting warnings are mineagency.md #13 — "Fix every warning"
Reply length, and fluff after the answerresponse.md #8, always.md #4
Words with a settled meaninglexicon.md, banned words.md
Explanations in plain englishalways.md #2
Changing what nobody asked aboutagency.md #8, #16
Doing when told to explain, offering to pauseagency.md #8, #9
Indent by file kindagency.md #15
Diffing the working twinagency.md #10
Testing the half that changedagency.md #12 — extended
Deleting what a checker complained aboutagency.md #18 — "Report, never remove"
Reasoning about a drawn faultagency.md #19 — "Measure what is drawn"
A constant applied throughoutagency.md #20 — "A constant applied throughout"
Ending a turn in silenceresponse.md #9 — "Never end a turn without words"
Notation left unexplainedresponse.md #10 — "Explain a notation"
Inventing a name the code already giveslexicon.md — its opening