Auto-Improve#

Reads logs, detects patterns, rewrites the playbook. Not passive logging — this ACTS on what it learns.

text

SCAN (read logs) ──► PROPOSE (specific edits) ──► APPLY (low-risk auto, high-risk flag)

Input Sources#

SourceWhat It Contains.learnings/ERRORS.mdWhat broke and how it was fixed.learnings/LEARNINGS.mdCorrections, insights, knowledge gaps, batch outcomesworkspace/OUTSTANDING.mdRanked ideas and opportunitiesmemory/permanent/*.mdCurrent knowledge stateworkspace/DELEGATION_PLAN.mdAtom timing data (if delegation was used)

Step 1: SCAN#

Detect:

• Repeated errors — same mistake 3+ times → needs a prevention rule
• Repeated corrections — user keeps fixing the same thing → behavior change needed
• Emerging patterns — 3+ items connecting → thesis forming
• Stale knowledge — facts in permanent memory contradicted by recent sessions
• Unused wins — high-value items that haven't been acted on

Step 2: PROPOSE#

For each detected pattern:

text

PROPOSAL: [short title]
EVIDENCE: [file#line references]
CHANGE: [exact edit — old text → new text]
RISK: [low/medium/high]
REVERSIBLE: [yes/no]
Pattern-Key: [hash(error+fix) for dedup]

Pattern TypeActionTarget FileRepeated errorAdd prevention rulerelevant skill's ## Learned sectionRepeated correctionUpdate behavior guidelineSOUL.md or AGENTS.mdEmerging thesisWrite thesis + next stepsOUTSTANDING.mdStale knowledgeUpdate the factmemory/permanent/*.mdUnused winCreate ticket or reminderNEXT_TICKET.md or cron

Step 3: APPLY#

• Low risk + reversible: Apply immediately. Log the change.
• Medium risk: Apply but notify user on next interaction.
• High risk: Write to OUTSTANDING.md and wait for approval.
• Dry-run mode ( --dry-run): Propose all changes but apply none. Output a report.

Use 3-occurrence threshold before proposing pattern-based changes. Track recurrence with Pattern-Key and Recurrence-Count.

Error→Skill Feedback Loop#

After SCAN, for each error in ERRORS.md:

1. Extract the Context column value

2. Match against skill names (fuzzy: "SiteBlitz CSS" → webdev-sop)

3. If match found and skill doesn't already have the fix in ## Learned: markdown

   ## Learned
   

• [date] [error summary] → [fix]. Source: .learnings/ERRORS.md#L[N]

4. Use `Pattern-Key: hash(error+fix)` to prevent duplicates

Skills self-heal: every failure improves the relevant skill.

## Delegation Feedback

After delegation plan completes:

1. Read atom timing data from DELEGATION\_PLAN.md
2. Atom actual time > 2× estimated → flag estimation drift
3. Atom model upgraded (flash→sonnet) → update routing suggestion in MODEL\_ROUTING\_PROTOCOL.md
4. Append summary to `.learnings/LEARNINGS.md`