Memory Ready Score

Public contract for memory-audit.

Same ritual as Loop Ready / Fleet Ready:

node tools/memory-audit/cli.js . --suggest
# later: npx @cobusgreyling/memory-audit . --suggest

Levels

Level	Score	Meaning	Safe behavior
M0	0–39	Ad-hoc memory	Chat history only; no shared policy
M1	40–64	Structured	Scratch + episodic files; no auto durable
M2	65–84	Budgeted	Recall caps + hygiene loop
M3	85–100	Gated durable	Verifier/human promotion + CI dogfood

CI examples:

node tools/memory-audit/cli.js . --min-score=40   # M1
node tools/memory-audit/cli.js . --min-score=65   # M2

Weights (v0)

Base 10.

Signal	Points
MEMORY.md posture	+16
MEMORY-STATE.md	+14
Tiers documented in MEMORY.md	+8
memory-write skill	+8
memory-recall skill	+8
memory-hygiene skill	+8
memory-verifier skill	+8
memory-budget.md	+10
memory-constraints.md	+6
memory-run-log.md	+4
Confidence tags in use	+4
GitHub audit workflow	+4
Patterns registry present	+2

Penalties:

Issue	Penalty
Durable section without any owner/evidence hints	−8
Memory files mention secrets patterns	−15
STATE.md > 100KB with no MEMORY split	−5

Human output template

Memory Ready: 47/100  ·  M1 structured

OK
  MEMORY.md present
  MEMORY-STATE.md present
  memory-write skill
  memory-recall skill

WARN
  Missing memory-budget.md
  Missing memory-hygiene skill

Suggested actions
  node tools/memory-init/cli.js . --pattern retrieval-budget
  node tools/memory-init/cli.js . --pattern memory-hygiene-loop

Climb path

1. memory-init --pattern session-scratchpad → ~M1
2. Add episodic + budget → mid M1 / low M2
3. Hygiene loop + verifier → M2
4. Durable store with gates + CI → M3