Tree Ring Memory

Tree Ring Memory

Tree Ring Memory is a framework-agnostic, local-first memory lifecycle layer for AI agents. Use this skill when an agent should preserve useful decisions, warnings, preferences, evidence, or future follow-ups without keeping raw conversation transcripts.

The public runtime is a Rust CLI with SQLite/FTS storage, recall, audit, forgetting, deterministic consolidation, JSONL import/export, DOX/Revolve source adapters, framework discovery, and a Ratatui terminal console.

When To Use

Use Tree Ring Memory before:

• starting or resuming work in a project with existing memory
• making architecture, privacy, security, release, or storage decisions
• repeating a workflow where prior mistakes may matter
• responding to a user correction or durable preference
• closing meaningful work and deciding what should help future agents

Use it after:

• a decision is validated by tests, review, or production behavior
• a failed approach should not be repeated
• a useful project convention is discovered
• the user explicitly says something should be remembered
• an evaluated outcome should become evidence, a warning, or a future seed

Memory Model

Tree Ring Memory treats memory as something that ages:

• cambium: active or recent task context
• outer: recent project lessons and decisions
• inner: older compressed project knowledge
• heartwood: durable high-confidence truths
• scar: important failures, regressions, rejected approaches, and warnings
• seed: unresolved hypotheses and future work

Prefer concise memories. Store the lesson, decision, warning, or evidence summary, not the full conversation.

Install Or Discover

If Tree Ring Memory is already initialized in the project, read:

.tree-ring/SKILL.md
.tree-ring/CLI.md

If the CLI is not installed, use one of the public install paths:

brew tap TerminallyLazy/tree-ring
brew install tree-ring

or:

curl -fsSL https://raw.githubusercontent.com/TerminallyLazy/Tree-Ring-Memory/main/install.sh | sh

Project repository:

https://github.com/TerminallyLazy/Tree-Ring-Memory

Core Commands

Inspect the local CLI first because preview commands can evolve:

tree-ring --help
tree-ring init --help
tree-ring remember --help
tree-ring recall --help
tree-ring evidence --help
tree-ring audit --help
tree-ring forget --help
tree-ring maintain --help

Typical local workflow:

tree-ring init
tree-ring remember "Use project-scoped recall before risky release changes." --event-type lesson --scope project
tree-ring recall "release changes"
tree-ring audit --stale-after-days 60

Use source adapters with dry runs first:

tree-ring dox sync --source-root . --dry-run
tree-ring revolve sync --source-root revolve --dry-run
tree-ring integrations scan --source-root .

Only write adapter summaries when they are concise, useful, and source-linked. Memory does not replace the source AGENTS.md, Revolve record, evaluation, issue, PR, or test artifact.

Evidence Records

Use tree-ring evidence instead of a plain memory when the lesson comes from:

• tests or benchmarks
• release validation
• an incident or rollback
• an evaluation run
• reviewed branch, PR, issue, or run artifact

Outcome mapping:

• promoted: durable heartwood from strong evidence
• rejected: scar for reusable failed or rolled-back approaches
• deferred: seed for promising unresolved options
• observed: outer-ring evaluation result

Do not promote weak or unreviewed claims to heartwood.

Privacy And Safety

Tree Ring Memory should not be used as a hidden recorder.

Do not store:

• secrets, credentials, tokens, or private keys
• raw chain-of-thought or private scratchpad content
• temporary notes with no future value
• unverified claims as durable truth
• sensitive personal data unless the user explicitly asks and the summary is safe to retain
• copyrighted source text beyond short allowed snippets

If a useful memory contains sensitive details, store a redacted summary with the minimum context needed for future work.

Closeout Habit

At the end of meaningful work, ask:

• What did we decide?
• What did we learn?
• What should future agents avoid repeating?
• Did the user state a durable preference?
• Is there a future seed worth revisiting?
• Is any memory sensitive and better left unstored?

Only write memories that will materially improve future work.