# workspace-metabolism

MCP server and CLI for governing files left by AI coding agents: policy-driven
audit, reversible cleanup, rollback, and hash-chained verification. Python
3.11+, **zero dependencies**, Windows / Linux / macOS.

[![PyPI version](https://img.shields.io/pypi/v/workspace-metabolism)](https://pypi.org/project/workspace-metabolism)
[![Python](https://img.shields.io/pypi/pyversions/workspace-metabolism)](https://pypi.org/project/workspace-metabolism)
[![CI](https://img.shields.io/github/actions/workflow/status/metabolism-tools/workspace-metabolism/ci.yml)](https://github.com/metabolism-tools/workspace-metabolism/actions)
[![License: MIT](https://img.shields.io/github/license/metabolism-tools/workspace-metabolism)](LICENSE)
[![Zero dependencies](https://img.shields.io/badge/dependencies-0-brightgreen)](pyproject.toml)
[![Glama score](https://glama.ai/mcp/servers/metabolism-tools/workspace-metabolism/badges/score.svg)](https://glama.ai/mcp/servers/metabolism-tools/workspace-metabolism)

![Terminal demo](docs/demo-terminal.gif)

![workspace health](https://img.shields.io/endpoint?url=https%3A%2F%2Fraw.githubusercontent.com%2Fmetabolism-tools%2Fworkspace-metabolism%2Fmain%2Fdocs%2Fhealth.json)

▶️ Watch the 60-second animated demo: [docs/demo-terminal.html](docs/demo-terminal.html)

## The problem

AI coding agents (Claude Code, Codex, DeepSeek Harness, …) share one thing —
your workspace — and they leave a trail of scratch files, caches and staged
directories behind. Nobody owns the cleanup: deleting by hand is irreversible,
scheduled scripts have no audit trail, and the next agent run works in the
garbage the last one left.

`workspace-metabolism` is the **policy layer** for that: one `metabolism.json`
decides what every path is worth (G1 never touch → G4 auto), nothing is ever
deleted by pattern — items move to a recycle area with per-file SHA-256 hashes
and `rollback` restores them exactly — and every action lands in a
**hash-chained journal** that `verify` can audit.

**Try it in 30 seconds:**

```bash
pip install workspace-metabolism
wm doctor --residue                  # what agent byproducts your policy doesn't govern yet
wm doctor --residue --apply-policy   # adopt the suggestions as policy entries (creates the file if missing)
wm audit                             # read-only checkup with health score
```

**Distribution (v0.7.0):** [GitHub release and installable wheel](https://github.com/metabolism-tools/workspace-metabolism/releases/tag/v0.7.0).
`retain` requires this GitHub wheel/source; PyPI publication is separate.
See [v0.7.0 release notes](docs/publish/release-notes-v0.7.0.md) for installation.
Listed in [awesome-mcp-servers](https://github.com/punkpeye/awesome-mcp-servers) under Developer Tools (merged 2026-09-21).
See the [Glama tool-definition assessment](https://glama.ai/mcp/servers/metabolism-tools/workspace-metabolism/score)
for interface quality; it does not establish production reliability or lower supervision.
Honest: no large production deployments yet, and the policy schema may shift
before v1.0. Early adopters are welcome to break it on weird directory
structures.

**Honest boundaries** — what this is *not*:

- **Not a sandbox.** `wm gate` is a governance/audit layer for cooperative
  agents; a compromised or malicious agent can bypass it and call the target
  server directly. OS-level sandboxing is a different layer.
- **Not a heuristic classifier.** It never decides "this file is garbage" on
  its own — only the policy you approved decides. `doctor` only *suggests*
  entries; nothing is governed until you adopt them.
- **Does not fix agent bugs.** It governs the byproducts agents leave; it does
  not stop agents from producing them.
- **Local audit, not a notary.** The hash-chained journal detects tampering
  with the tool's own records; it is not a distributed or court-grade ledger.

This repo has two linked ideas:

- **Agentic Metabolic Engineering** is the method: how to think about workspace lifecycle.
- **AI governance as code** is the implementation: how `wm` applies that method with policy files and commands.

## 中文快速上手（30 秒）

**v0.6.0 新增**：先认领再写入、清理避让未完成认领、精确绑定 SQLite 路径及不建库的只读
检查。复用现有任务和数据入口，不另建 AI 调度系统。见[接入指南](docs/existing-system-integration.md)、
[认领使用说明](docs/claim-before-write.md)和[数据库检查](docs/sqlite-resources.md)。

AI 编程（Claude Code / Codex / Aider 等）会在工作区留下大量草稿、缓存和
废弃文件，越堆越多，下一轮 AI 还得在垃圾堆里干活。这个工具用一份策略文件
管理文件的整个生命周期：**检查（只读）→ 回收（可回滚）→ 验证（防篡改记录）
→ 清理**。

```bash
pip install workspace-metabolism        # 安装（零依赖）
python examples/demo.py                 # 30 秒演示：盲删 vs 回收+回滚
wm init                                 # 生成策略文件 metabolism.json
wm audit                                # 只读体检，给文件贴营养标签
wm clean --grades G4 --yes              # 回收过期项（默认 dry-run，确认后加 --yes）
wm rollback <run_id>                    # 删错了？一键原样找回
wm govern write --path src/main.py      # 写文件前先问策略：允许吗？（AI 执行点拦截）
wm slim --db data/app.db --yes          # 数据库也会膨胀：策略驱动的库内瘦身（v0.3）
```

默认只读、绝不直接删文件；每步操作都有防篡改记录；Windows / Mac / Linux 通用。
项目处于早期，认领和受控编辑仍为实验能力；策略格式在 v1.0 前可能调整。完整英文文档见下文。

## Why this exists

Most disk tools either show you space (`ncdu`, `duf`) or delete things
(`rmlint`). `workspace-metabolism` is different: a **policy file** defines what
every path is worth (grades G1–G4), and the tool only ever does what the policy
allows — nothing more. It is the **policy layer for multi-agent workspaces**:
Claude Code, Codex, Aider, OpenClaw and every other agent share one thing —
your workspace — and the policy governs the byproducts all of them leave
behind, regardless of which tool created them. It fixes no vendor and judges
no file; see [What this is not](docs/positioning.md) before you judge it.

- **G1 never touch** / **G2 keep** / **G3 approve + reference check** / **G4 auto**
- Deletion is never direct: items move to a recycle area, then `rollback`
  restores them after a per-file SHA-256 integrity check
- Every action lands in a **hash-chained journal**; `verify` detects any edit
- Read-only `audit` reports candidates, unregistered paths, disk alerts, growth
  trend and possible duplicates — plus residue on **memory-backed mounts**
  (tmpfs/ramfs: it costs RAM, not just disk)
- Optional **protected window** (e.g. trading hours, business hours) during
  which marked entries are never touched
- Scheduled runs are supported out of the box on Windows (Task Scheduler) and
  Linux/macOS (cron) via templates in `examples/`

### Why not just a scheduled cleanup?

A scheduled task — or asking Codex to "clean up old files" on a timer — gets
you *at some point, files get removed*. `workspace-metabolism` gets you:

- rules that live in the repo (`metabolism.json`), versioned and reviewable
- cleanup that never deletes directly: recycle area, per-file SHA-256, exact
  `rollback`
- a hash-chained journal that detects tampering
- the same behavior on every machine and every run, no AI judgment involved

Scheduling and metabolism are complementary, not rivals: this repo ships cron,
Windows Task Scheduler and CI templates that run `wm` itself. The scheduler
answers *when*; the policy answers *what, how, and how to undo it*.

### What this is not

Four objections come up so often they deserve their own page
([docs/positioning.md](docs/positioning.md)). The short version:

- **Not a fix for vendor bugs** — Claude Code's `/tmp` leak, OpenClaw's
  staged-dir residue: those belong upstream. We govern the workspace, which
  is the one thing every agent shares.
- **Not a heuristic classifier** — no guessing, no AI judgment. Only the
  policy file you wrote decides anything; `wm explain <path>` shows the rule.
- **Not a rival to agent self-cleanup** — agents should clean up after
  themselves; `wm mcp` + session-end hooks make that safe and audited.
- **Not a blind-delete script** — nothing is ever deleted by pattern: items
  move to a recycle area with per-file hashes, and `rollback` restores them.
  `purge` deletes file-recycle batches. The explicit `retain` policy can remove
  unreferenced database rows after writing a complete recovery export.

## See it in action

This repo ships a reproducible benchmark: two identical workspaces run 30
simulated agent loops; one ends every loop with `wm clean`, the other never
cleans. The result — **2 active files vs 242** — is a number you can reproduce
yourself:

```bash
python examples/metabolism_benchmark.py
```

A recorded run (2026-08-16, wm 0.2.0) is in
[docs/publish/benchmark-run-20260816.json](docs/publish/benchmark-run-20260816.json)
(raw log:
[docs/publish/benchmark-run-20260816.txt](docs/publish/benchmark-run-20260816.txt)).

### Case study: a 20.7 GB database that stalled a research engine

`wm slim` was born from a production incident, and the dogfooding round
produced the most honest review the tool has had. Read
[docs/case-studies/research-engine-db-rot.md](docs/case-studies/research-engine-db-rot.md):
three failure modes (dead work units, database rot, silently-dead jobs), the
fixes, and what we found when we used `wm slim` to verify them — a policy
stripping the wrong field, two path-matching bugs, a CLI flag-order pitfall
that failed the first scheduled run, and the first successful run reclaiming
**10.15 GB** (21.7 GB → 11.3 GB) before uncovering a third real problem: the
dead-position exclusion rule forgot itself once "clean" epochs diluted its
learning window. Real usage is the final test.

## 🧬 Philosophy

`workspace-metabolism` treats your AI-generated workspace as a finite system:
audit → clean → verify → rollback, with recyclable cleanup and a hash-chained
audit trail. Cleanup is the means; metabolism is the frame. The one-liner:
**loops keep the agent running; metabolism keeps the workspace usable.** We
call this framing **Agentic Metabolic Engineering** — managing the byproducts
of agent-driven software workspaces. Full write-up:
[docs/philosophy.md](docs/philosophy.md) · [the story](docs/narrative.md) ·
[competitive analysis](docs/competitive-analysis.md) ·
[academic anchors](docs/academic-anchors.md).

## Quick start

```bash
# install the release that includes retain (PyPI is a separate channel)
pip install https://github.com/metabolism-tools/workspace-metabolism/releases/download/v0.7.0/workspace_metabolism-0.7.0-py3-none-any.whl

# or run without installing anything:
#   PYTHONPATH=src python -m workspace_metabolism --help

# try it on a throwaway workspace (builds demo files; shows the usual
# blind-delete fix vs the wm way: recycle + rollback + journal)
python examples/demo.py
```

Point the tool at your own workspace:

```bash
cd /path/to/workspace
wm init            # scaffold metabolism.json (like `git init`)
wm doctor          # check readiness before the first audit or cleanup
wm audit           # first checkup (read-only)
wm health          # workspace health score (0-100)
wm explain logs    # why a path is graded the way it is
wm clean --grades G4 --yes   # recycle expired G4 items (dry-run without --yes)
wm rollback <run_id>
```

`wm init` scans your workspace and registers common directories (source and
docs as G2 keep, logs/tmp/cache as G4 auto, archive/staging as G3 approve).
Edit `metabolism.json` and commit it like any source file. The tool
auto-discovers `metabolism.json` (or `.wm.json`) in the workspace root, so
`--registry` is optional. Nothing is cleaned unless it is registered in the
policy file. Advanced users can start from
[examples/registry.example.json](examples/registry.example.json).

## Commands

| Command | What it does |
| --- | --- |
| `audit` | Read-only health check; writes a report and a journal entry (also flags sensitive files and git-tracked content) |
| `clean --grades G4` | Move expired items to the recycle area (dry-run by default) |
| `clean --grades G3` | Same, but requires `--approve` + `--approver` |
| `rollback <run_id>` | Restore one cleanup run after an integrity check |
| `purge --older-than 30` | Delete expired file-recycle batches |
| `retain --db data/objects.db` | Preview removal of excess, unreferenced row versions under an explicit policy |
| `retain --db data/objects.db --restore RUN_ID` | Preview verified row restoration; add `--yes` to execute |
| `verify` | Check the journal hash chain and run manifests |
| `status` | Overview of workspace, recycle area and pending candidates |
| `init` | Scaffold a `metabolism.json` policy file (like `git init`) |
| `explain <path>` | Show what the policy says about a path (the nutrition label) |
| `health` | Workspace health score (0-100), with `--json` and `--badge` output |
| `doctor` | Read-only readiness check (workspace, policy, state, locks); `--residue` also lists common agent byproducts (`.cursor`, `.claude`, caches, logs) the policy does not govern yet — each with the exact policy entry that would govern it, `--apply-policy` adopts them |
| `govern <action>` | Check whether an AI action is allowed by policy and record the decision |
| `gate --target ...` | MCP governance proxy: every tool call of the wrapped server is checked against the policy first |
| `slim --db PATH` | Policy-driven in-place trimming of heavy JSON fields in a SQLite database (journaled; dry-run by default) |
| `mcp` | MCP stdio server so agents can run micro-metabolism themselves |

Global flags:

| Flag | Meaning |
| --- | --- |
| `--root PATH` | Workspace to govern (default: current directory) |
| `--state-dir PATH` | Journal / recycle / runs / reports (default: system cache directory, **outside** the workspace) |
| `--registry PATH` | Policy JSON (optional; auto-discovers `metabolism.json` / `.wm.json`) |
| `--protected-window HH:MM-HH:MM` | Weekday window; entries marked `protected` are skipped while active |

The default state directory lives outside the workspace on purpose — a
`git add .` in your project can never sweep the audit journal into version
control.

`wm doctor` is a read-only preflight check. It reports whether the workspace
and state directory are writable, whether the policy exists and is valid, and
whether another `wm` operation currently holds the state lock. The lock
serializes audits, cleanup, rollback and purge so concurrent scheduled or
agent-triggered runs cannot interleave journal and recycle operations.

### AI governance as code

The optional `ai_governance` section is the concrete implementation of this
repository's AI governance layer. It uses the same policy file to check AI
actions before they happen. Unknown actions are denied by default; write
actions can require a preview, while execute, delete and network actions can
require a named approver. `wm govern` only makes and records a decision; it
does not perform the action for the caller.

```bash
wm govern write --path src/main.py
wm govern write --path src/main.py --preview
wm govern execute --path scripts/release.ps1 --approve-by "name"
wm govern network --approve-by "name" --json
```

**`wm gate` turns decisions into enforcement.** It wraps any MCP stdio server
and checks every `tools/call` against the policy before forwarding it;
denied calls never reach the target and every decision lands in the journal:

```bash
wm gate --target "python -m my_mcp_server"
```

Map tool names to actions with `tool_patterns` (glob), e.g.
`"fs_write*": "write"`, `"shell*": "execute"`. Unmatched tools default to the
`execute` action. For tools whose calls carry a preview mode, pass
`"preview": true` in the call arguments to satisfy `requires_preview`.

Every decision includes the policy hash and is written to the same
hash-chained journal; `govern` returns a `decision_id` that `clean` /
`rollback` / `slim` accept via `--decision-id`, so the journal shows the full
**intent → decision → execution** chain. The approver value is an auditable
declaration, not an authentication mechanism.

> **Honest boundary:** `wm gate` is a governance and audit layer, **not a
> sandbox**. A compromised or malicious agent can bypass the proxy and talk
> to the target directly. Gate governs the cooperative agent; OS-level
> sandboxing governs the hostile one.

**First run, guided:** `wm doctor --residue` scans for the byproducts agents
usually leave behind (`.cursor`, `.claude`, `node_modules/.cache`,
`__pycache__`, logs …) that your policy does not govern yet. Every hit shows
the exact policy entry that would govern it; `--apply-policy` adopts the
suggestions into `metabolism.json` (creating it if needed). Nothing is ever
deleted — the suggestions become policy, and the policy still decides
everything afterwards:

```bash
wm doctor --residue               # what is ungoverned, and the suggested entries
wm doctor --residue --apply-policy  # adopt them as policy entries, then audit
```

## Policy file

```json
{
  "version": 1,
  "defaults": {
    "recycle_retention_days": 30,
    "max_item_mb": 2560,
    "disk_alert_free_gb": 20,
    "disk_alert_free_pct": 15,
    "dupe_scan_dirs": ["tmp", "cache"]
  },
  "never_clean": [".git", "README.md", "src"],
  "entries": [
    {"path": "logs", "grade": "G4", "cleanup": "auto", "retention_days": 30},
    {"path": "archive", "grade": "G3", "cleanup": "approve", "retention_days": 60},
    {"path": "**/__pycache__", "grade": "G4", "cleanup": "auto", "retention_days": 30}
  ]
}
```

| Field | Meaning |
| --- | --- |
| `path` | Path or glob (`*`, `**/`) relative to `--root` |
| `grade` | G1 never / G2 keep / G3 approve / G4 auto |
| `cleanup` | `never`, `auto` or `approve` |
| `retention_days` | Idle days before the item becomes a candidate (required unless `cleanup=never`) |
| `scope` | Optional: `files_only` (top-level files of a directory) |
| `protected` | Optional: skip while a `--protected-window` is active |
| `remote_authoritative` | Optional: display marker for data with a remote source of truth |
| `category` | Optional free-form label for your own classification |
| `owner` | Optional: who is accountable for this rule |
| `intent` | Optional: why this rule exists |
| `review_after` | Optional: when this rule should be revisited |

The policy format is versioned and validated against
[schema/metabolism.schema.json](schema/metabolism.schema.json), so editors and
agents can check your file before the tool does.

## Health score

`wm health` combines the audit summary into one number from 0 to 100: 25
points for journal auditability, 25 for governance (unregistered paths, disk
alerts), 35 for rot burden (expired candidates), and 15 for recycle
readiness. Grades: A (90+), B (75+), C (60+), D (below).

```bash
wm health --json
wm health --badge   # shields.io endpoint JSON for a README badge
```

The badge above is generated from [docs/health.json](docs/health.json). A CI
template that fails when the score drops below a threshold is in
[examples/ci-audit.yml](examples/ci-audit.yml).

## Agents

`wm mcp` runs a zero-dependency MCP stdio server. Agents can init a policy,
audit, explain, verify, and dry-run clean plans themselves; `clean` only
executes when the caller explicitly passes `execute=true`, `rollback` restores
a previous run from the recycle area (SHA-256 verified), and the policy file
still decides everything. The end-of-loop ritual is automated in
[examples/micro_metabolism.py](examples/micro_metabolism.py) — wire it into a
session-end hook so every loop ends with a checkup.

### DeepSeek Harness (DSH)

DSH is an agent harness where *everything is a plugin* (Cordis). Its official
third-party tool channel is MCP, and `wm mcp` already speaks it — one
`cordis.yml` row exposes all nine wm tools to the DSH agent (audit, health,
explain, verify, `wm_db_check` registered-SQLite checks, `wm_govern` pre-action
policy checks, clean, init, rollback):

```yaml
- insert:
    - id: workspace-metabolism
      name: '@deepseek-ai/dsh-mcp-client'
      config:
        serverName: wm
        transport: stdio
        command: wm
        args: [mcp]
        cwd: !!js process.cwd()
```

Full walkthrough (project `cordis.yml` vs `--patch` overlay, pinned
`--root`/`--state-dir`, safety notes):
[docs/dsh-integration.md](docs/dsh-integration.md). A policy tuned for
DSH-style workspaces (`.agents/notes`, scratch plugins, generated artifacts):
[examples/registry.dsh.example.json](examples/registry.dsh.example.json).

The optional [Metabolic Maintenance skill plugin](plugins/dsh-metabolic-maintenance/README.md)
adds consequence-aware maintenance rules and downstream recovery verification to
DSH's skill catalog. It loads instructions on demand and can pair with the MCP
tools above. [Download the plugin](https://github.com/metabolism-tools/workspace-metabolism/releases/tag/dsh-metabolic-maintenance-v0.1.3).
Releases are also mirrored to the plugin's distribution repository,
[metabolism-tools/dsh-metabolic-maintenance](https://github.com/metabolism-tools/dsh-metabolic-maintenance),
which carries the `dsh-plugin` topic so plugin directories can index it as a plugin.

## Safety model

- `clean` is dry-run unless `--yes` is given.
- G4 needs `--yes`; G3 needs `--approve` **and** `--approver` (audit trail).
- **Sensitive files are never auto-cleaned**: `audit` flags secrets/keys/credentials
  (`.env*`, `*.pem`, `*.key`, `*token*`, `*secret*`, `*credential*`, `id_rsa`, …) in a
  dedicated report section, the policy validator refuses to register a sensitive path
  as G4 auto-clean, and `clean` skips any candidate that contains sensitive files.
- **Git-aware classification**: in a git repo, tracked files count as controlled by git
  (effectively G2) — they are excluded from the audit's unregistered list, and `clean`
  skips candidates that contain git-tracked files. Non-git workspaces fall back to pure
  policy matching. (Git is optional; `wm` never depends on it.)
- Items move to the recycle area with per-file SHA-256 hashes; `rollback`
  verifies them before restoring and refuses to overwrite an existing path.
- `purge` deletes file batches only inside the recycle
  area after retention.
- The journal is a hash chain; `verify` detects any tampering.

## Scheduled runs

Templates with `{{PLACEHOLDERS}}` are in `examples/`:

- Windows — `register_schedule.template.ps1`: daily read-only audit (20:30),
  weekly G4 clean (Saturday 10:00), monthly purge (1st, 10:30).
- Linux/macOS — `register_cron.template.sh`: same schedule via cron.

Replace `{{WM_CMD}}`, `{{ROOT}}`, `{{REGISTRY}}`, `{{STATE_DIR}}` (and
`{{USER}}` in cron) with your values. The scripts deliberately do **not**
auto-detect your environment — your paths, your call.

## Development

```bash
python -m pip install -e . pytest
python -m pytest
```

CI runs the full test suite on Ubuntu, Windows and macOS with Python 3.11 and
3.12. Issues are handled on weekends; pull requests are welcome.

## Project family

Sister organization: [Holdout](https://github.com/holdout-labs) — a
toolchain against self-deception in quantitative research:

- [`falsification-ledger`](https://github.com/holdout-labs/falsification-ledger) — pre-registration and falsification ledger
- [`factor-qc`](https://github.com/holdout-labs/factor-qc) — fail-closed backtest quality gate
- [`pit-adjuster`](https://github.com/holdout-labs/pit-adjuster) — PIT back-adjustment with drift detection
- [`lookahead-free`](https://github.com/holdout-labs/lookahead-free) — verifiable look-ahead-freedom checks
- [`ashare-data-immunity`](https://github.com/holdout-labs/ashare-data-immunity) — data immunity for A-share daily bars
- [`lesson-book`](https://github.com/holdout-labs/lesson-book) — tuition memory for traders

If `workspace-metabolism` keeps the *workspace* alive, Holdout keeps
the *research* honest.

## License

MIT

## SQLite row retention

`retain` keeps the newest N versions per JSON group **plus every referenced
version**, including references from old payloads. It never changes blob contents.
The most specific workspace-relative entry must explicitly allow `db_retain`
on a G2 database; G1, `never_clean`, active claims and `--protected-window` block
execution. There are no CLI overrides that weaken the policy.

```json
{
  "path": "data/objects.db", "grade": "G2", "cleanup": "never",
  "db_retain": {
    "table": "objects", "id_column": "hash", "blob_column": "body",
    "group_by": ["code", "month"], "order_column": "captured_at", "keep": 2,
    "blob_encoding": "gzip", "verify_sha256": true,
    "reference_paths": ["objects/*/*"]
  }
}
```

Here `objects/*/*` traverses a JSON dictionary of lists of object IDs in **all**
rows. Missing optional fields contain no references; malformed containers,
missing referenced rows, invalid JSON, hash mismatches, unsupported schema or
exceeded budgets stop the operation. Ungrouped payload rows are always retained.
An empty `reference_paths` explicitly asserts that rows have no in-table
references. WM cannot discover undeclared or external consumers; policy authors
must establish that boundary before authorizing deletion. Triggers, foreign-key
relationships, generated columns and composite primary keys are not supported.

```bash
wm --root /workspace --registry policy.json retain --db data/objects.db
wm --root /workspace --registry policy.json retain --db data/objects.db --yes
wm --root /workspace --registry policy.json retain --db data/objects.db --restore retain-RUN_ID
wm --root /workspace --registry policy.json retain --db data/objects.db --restore retain-RUN_ID --yes
```

Use the actual `run_id` returned by the executed command. Preview writes no
maintenance state. Execution plans inside a SQLite write transaction, exports
**all columns with their types**, persists and reads back the export, then deletes.
Recovery lives in `<state-dir>/retention/<run_id>/`, outside file `purge`'s scope,
with no automatic expiry. Keep it until downstream acceptance and recovery
requirements are satisfied; do not reclaim these backups by age alone.
An interrupted `prepared` batch blocks another retention run; inspect it and use
`--restore` to reconcile it. Restoration verifies the export, original database
identity and schema, and refuses to overwrite changed rows. Already identical
rows are left alone, making an interrupted restore retryable. Hashes detect
accidental changes, not malicious replacement of both export and manifest.

Defaults cap scans at 1,000,000 rows / 512 MiB encoded content / 120 seconds,
decoded rows at 32 MiB and recovery export at 64 MiB. Policy can lower these
limits. There is no automatic VACUUM, so reports claim zero disk reclamation;
deleted database pages remain reusable inside SQLite, and recovery files consume
space. Storage checks do not establish downstream business correctness.
See `tests/test_retain.py` for reference protection, failure injection, concurrent
writer exclusion, exact restoration and refusal of corrupted recovery material.

Trusted host integrations may supply a claim backend implementing
`transaction()` and `cleanup_claims()` to the Python `retain` function. The
backend must reuse the host's authoritative lock and retain active/unresolved
scopes; the CLI never ignores a foreign registry. `eligible_ids` can only narrow
WM's own safe candidates, and `expected_ids_sha256` rejects a stale caller
snapshot while the SQLite transaction is held.

## Maintenance evidence (0.5.1)

`wm --state-dir /path/to/state evidence` prints a bounded, read-only JSON
summary of `journal.jsonl`. Optionally add `--observation check.json` to read
an existing check containing timezone-aware `checked_at` and boolean `ok`.
No policy is required. Each input is limited to 8 MiB; no files are written.

Exit 0 means a nonempty internally consistent journal and, when requested,
a readable valid check. It does **not** mean the system is healthy. Missing,
empty, damaged, unsupported or oversized evidence returns exit 1.
Time association does not establish matching scope, freshness, causality or
business recovery. Human supervision, Token cost and net savings remain unknown.
See [case collection guide](docs/case-studies/README.md).

## Safe recent-row retention (0.5.2)

`slim` can now match a real database relationship instead of assuming the
ordering timestamp is embedded in every JSON payload:

```json
"db_slim": {
  "table": "work_units",
  "blob_column": "payload_json",
  "strip_keys": ["regenerable_detail"],
  "protected_keys": ["consumer_evidence", "failure_reason"],
  "keep_recent": {
    "table": "epochs", "column": "created_at", "n": 3,
    "key_column": "epoch_id", "row_column": "epoch_id"
  },
  "vacuum_min_gb": 1.0
}
```

`column` sorts the reference table; `key_column` joins to the work table's
`row_column`. Reference keys must be unique and non-null. For a row that would
change, a missing relationship stops the whole plan before any update. The
report's `rows_kept_recent` counts otherwise-modifiable rows protected by age.
Policies that omit the two new columns retain legacy JSON matching, but missing
or unknown references now stop instead of silently stripping the row. Migrate
such policies before scheduled use. Protected keys are top-level JSON keys;
conflicting CLI strip requests are rejected. This is not recursive field matching.

Execution holds the wm state lock and a SQLite write transaction across planning
and updates. Preview opens the database read-only but still writes the existing
wm journal. External writers must still observe the application's maintenance
window; these locks do not validate consumer correctness or coordinate other
systems. Backup/restore and post-maintenance consumer checks remain required.