Project Memory Ledger (OpenClaw Skill)

AI-ready engineering project ledger for long-term memory + evidence chains (Invariants / Decisions / Changes) with traceability and rollback.

Keywords: project ledger, engineering ledger, change log, decision log, invariants, traceability, rollback, evidence chain, agent memory, long-term memory, LLM ops, OpenClaw, Google Drive, Markdown

What it does

1) Local-first ledger (always local Markdown)

The ledger is always written to local Markdown files:

• <workspace>/ledgers/INVARIANTS.md
• <workspace>/ledgers/DECISIONS.md
• <workspace>/ledgers/CHANGES.md
• <workspace>/ledgers/PROJECTS.md
• <workspace>/ledgers/PRD_PATCHES.md

2) Project scaffolding (docs location controlled by backend)

backend decides where project docs/assets live:

• backend=local: create local project folders + Markdown docs
• backend=drive: create Drive project folders + GDoc/GSheet docs (requires gws authenticated)

This skill does not dual-write docs by design (avoids inconsistency). The ledger remains local.

---

Quick start

0) Choose a workspace

“workspace” means your OpenClaw working directory (not necessarily workspace-nero).

1) Use the ledger skill (write)

Configure and write L1 ledgers with project-memory-ledger.

2) Use the index skill (retrieve)

Build/query the L2 index with longterm-memory-index.

Example:

python3 longterm-memory-index/scripts/index_build.py \
  --workspace <workspace> \
  --db <workspace>/data/ledger_index.sqlite

python3 longterm-memory-index/scripts/index_query.py \
  --db <workspace>/data/ledger_index.sqlite \
  --q "keyword engine" \
  --limit 5

0) Choose a workspace

“workspace” means your OpenClaw working directory (not necessarily workspace-nero).

1) Create config

Recommended path:

• <workspace>/config/project_memory_ledger.json

Minimal config (local docs):

{
  "language": "en",
  "backend": "local",
  "default_project": "project-memory-ledger",
  "local_dir": "./ledgers",
  "projects_root_dir": "./projects",
  "projects_root_folder_id": ""
}

Drive docs config:

{
  "language": "en",
  "backend": "drive",
  "default_project": "project-memory-ledger",
  "local_dir": "./ledgers",
  "projects_root_dir": "./projects",
  "projects_root_folder_id": "<Drive folder id where projects should be created>"
}

Template:

• references/default_config.json

2) Init

python3 skills/project-memory-ledger/scripts/ledger.py init --config <workspace>/config/project_memory_ledger.json

3) Append entries

python3 skills/project-memory-ledger/scripts/ledger.py append \
  --config <workspace>/config/project_memory_ledger.json \
  --doc changes \
  --project "Keyword Engine" \
  --text "- **Interfaces**: hunter-system ↔ keyword-engine\n- **Change**: ...\n- **Why**: ...\n- **Evidence**: ...\n- **Rollback**: ...\n"

4) Register a project (scaffold)

python3 skills/project-memory-ledger/scripts/ledger.py register-project \
  --config <workspace>/config/project_memory_ledger.json \
  --name "My New Project" \
  --purpose "Why it exists + success criteria" \
  --interfaces "hunter-system ↔ keyword-engine" \
  --notes "optional"

• backend=local → creates <workspace>/projects/My New Project/... and Markdown docs
• backend=drive → creates Drive folders + GDoc/GSheet docs

5) Update PRD (propose-only)

python3 skills/project-memory-ledger/scripts/ledger.py update-prd \
  --config <workspace>/config/project_memory_ledger.json \
  --project "Keyword Engine" \
  --mode propose

---

Conventions

Project tagging

Project values are case-insensitive and normalized to slugs.

Recommended slugs:

• hunter-system
• keyword-engine
• common (cross-project rules/methods; not global AGENTS/MEMORY)

shared is disabled. Cross-project coupling must be explicit:

• Interfaces: a ↔ b

Default project

If you omit --project, the script injects:

• Project: project-memory-ledger

---

Drive backend options

Drive mode is optional and pluggable:

• Option 1 (v1 implemented): gws CLI (installed + authenticated)
• Option 2: Google API / service account
• Option 3: third-party Drive tooling

If Drive isn’t available, use backend=local. / service account

• Option 3: third-party Drive tooling

If Drive isn’t available, use backend=local.