v0.9.0

Fixed

• undo apply replays against the target the original write ran on. It dispatched the inverse against whatever target the caller named — in practice the config's first entry — while the write's own target sat unused in the undo record. On a multi-target config the inverse therefore ran against the wrong host; it only looks harmless because the resource usually is not there, but two hosts holding the same name and the inverse succeeds on the wrong one, silently. An explicitly named target still wins. Line-wide: all 24 copies had the identical defect. Caught live in container-host-aiops, where a stop recorded against a Podman target replayed against a Portainer one.

v0.8.0

Changed (BREAKING)

• Requires MCP SDK 2.0 (mcp[cli]>=2.0,<3.0). mcp.server.fastmcp no longer exists in 2.0; the server is now built with MCPServer and reports its package version in the stdio handshake.

Fixed

• undo apply works from the CLI. Every write tool is imported lazily inside its own CLI command, so a CLI-driven undo ran in a process where the inverse tool was never registered and failed with "inverse tool is not registered" — for every write tool. Only the MCP entry point, which imports the whole server, worked. Found while live-verifying against a real cluster.
• An undetermined outcome is audited unknown, not ok. The harness only classified a result as undetermined when the payload also carried an error key, so a write that looked successful but had not been confirmed was recorded as a success.
• OSD host leaked a Python dict repr and crushWeight was always null. Live-verified against Ceph 18 (reef): /api/osd returns host as a CRUSH bucket dict, not a string, and crush_weight is nested under tree. Both now read the real shape.
• crushWeight falls back to the tree only when the key is truly absent. .get(k, default) does not fire its default for a key that exists with a null value, so a build sending crush_weight: null at the top level would have reported null while the real number sat in tree. Uses is not None rather than truthiness, so an explicit 0 — a real CRUSH weight for a draining OSD — survives.

v0.7.0

Release notes — ceph-aiops 0.7.0

Previous release: 0.6.0.

Preview fidelity

A --dry-run should run the same guards as the real call and leave an audit row — the line's invariant is "a dry_run MAY read; it must never write." A few write commands still showed a hand-written banner that ran no guard and audited nothing. Those are now routed through the governed twin. The real writes were always guarded and audited; only the previews were blind.

In this tool

• osd reweight --dry-run now routes through the governed twin (reads the current + target weight, audits) instead of a static banner.

v0.6.0

Release notes — ceph-aiops 0.6.0

Previous release: 0.5.0.

BREAKING — the authorization layer is removed

This tool no longer decides whether a write is permitted. Read-only mode
(<PREFIX>_READ_ONLY), the graduated-approval / approver gate, and the
rules.yaml deny engine are all gone. Whether an operation runs is the
agent's judgement, or the permission of the account you connect it with — point
it at a read-only credential and the write fails at the server, the place that
actually owns the permission.

What the tool guarantees instead is that nothing is silent: every operation,
over MCP and the CLI alike, lands a row in the audit log — there is no
unaudited entry point. Destructive writes still capture their before-state and
record an undo token where a clean inverse exists.

• If you set <PREFIX>_READ_ONLY=1, it now has no effect and the MCP server
  logs a warning at startup. Restrict writes via the connecting account instead.
• <PREFIX>_AUDIT_APPROVED_BY / <PREFIX>_AUDIT_RATIONALE still work, but are now
  optional audit annotations — recorded on the row when set, never required.
• The declared risk_level is carried into the audit row as a descriptive tier
  (a label, not a gate).

The governance harness is now: audit (MCP+CLI, unbypassable) · runaway/budget
safety guard · undo recording · output sanitize. policy.py is a small
risk-tier classifier; governance/readonly.py is deleted.

Also in this tool

• set_pool_size refuses shrinking .mgr / .rgw.root — dropping the mgr pool to size 1 degrades ceph-mgr, which serves the Dashboard API this tool depends on. An increase is allowed.
• New scheme: (default https).

v0.5.0

Release notes — ceph-aiops 0.5.0

Previous release: 0.4.1.

In this tool

• delete_pool refuses .mgr and .rgw.root, and any pool whose application metadata marks it a mgr pool. Deleting .mgr breaks ceph-mgr, which serves the Dashboard API this tool's every call depends on.
• New scheme: (default https) — the base URL was hardcoded.

Every tool in the line: previews and undetermined outcomes

This release fixes three harness defects that were silently degrading the audit
trail and the undo store.

A write that loses its response is no longer recorded as a failure. The
harness assumed a sanitized error meant nothing had happened. That assumption is
false in exactly the case that matters most: when a write severs its own
connection, the request has already landed, the response cannot come back, and
the operation was recorded as status=error with no undo token created at
all. Transport-level failures are now audited as status=unknown, the result
says plainly that the operation may have taken effect and should be verified
before retrying, and a write that stashed its before-state has its inverse
recorded anyway — flagged effectVerified: false, which undo_list and
undo_apply both surface. Existing undo.db files are migrated in place; their
rows read as verified, which is accurate, since the old code only ever recorded
on the confirmed path.

A dry-run no longer writes an undo token. Previews were recording inverses
built from a before-state they never had: the undo callback's permissive default
filled the gap with a guess, producing a real, applicable token for an operation
that never happened.

A dry-run no longer demands a named approver. Requiring an approval in order
to ask whether something needs approval inverts what a preview is for. The tier
is still computed and still audited, so the preview can tell you an approver
will be needed; it just no longer refuses to answer. The write itself is gated
exactly as before.

The invariant, now stated: a dry_run may read; it must never write. Guards
run on the preview path, which means a preview can and does report that an
operation would be refused.

Also line-wide

• Truncated text now ends in an ellipsis instead of being cut silently. This
  line already treats a silent cut as a defect for lists; it was doing exactly
  that to strings.
• Error messages are capped at 800 characters, not 300. These messages end
  with what to do instead, so the cap was removing the most useful sentence of
  every long refusal.

v0.4.1

Release notes — ceph-aiops 0.4.1

Previous release: 0.4.0.

Fixed: pool_df reported a failed query as "no pools"

It caught every exception and returned an empty list, so a flaky mgr was
indistinguishable from a cluster that genuinely has no pools — the failure read as
"nothing to worry about".

BREAKING — pool_df now returns an envelope:
{"pools": [...], "returned": N, "error": str | None}. A non-null error means the
query failed. (The MCP wrapper already declared @tool_errors("dict"); its return
annotation now matches.)

This is the same defect class found and fixed in minio-aiops 0.3.0, where an
identical except: return [] hid a much larger bug for the life of the tool. A
probe failure must never be presentable as health.

v0.4.0

Release notes — ceph-aiops 0.4.0

Previous release: 0.3.0.

Headline: read-only mode

export CEPH_READ_ONLY=1

With this set the 19 write tools are never registered — an MCP
client lists 18 tools instead of 37. The writes are not hidden
behind a flag and not merely refused on call: they are absent from the session,
so a model cannot invoke one and cannot be argued into one. For a reviewer this
is checkable rather than promised — connect, list the tools, and the writes are
not there.

Enforcement is two layers deep: the @governed_tool harness refuses every
non-read operation (covering the CLI and in-process callers too), and the MCP
server removes write tools from list_tools(). Changing entry point does not
get around it.

Security fix included in this release

4 tool(s) documented as writes were carrying risk_level="low":
trigger_scrub, trigger_deep_scrub, cluster_flag_set, rbd_snapshot_create.

Because the read/write split keys off risk_level, read-only mode would have
left them exposed and able to execute real writes. They are now medium,
and a new test asserts risk_level can never again disagree with a tool's own
[READ]/[WRITE] documentation.

BREAKING — return shapes changed

This release changes payloads that callers may be parsing. All three changes exist
to stop a result from misrepresenting itself:

1. Absent fields are now null, not "". A missing value and an empty value
   were previously indistinguishable, which invited consumers to invent the
   difference. Keys are still always present — only the value may be null.
2. Anything with a limit now returns an envelope —
   {"<items>": [...], "returned": N, "limit": L, "truncated": bool}. Truncation is
   measured (one extra row is fetched), never inferred from the page happening to
   be full. Where a genuine pre-cap total is knowable it is reported as total;
   where it isn't, total is deliberately omitted rather than echoing returned.
3. risk_level changed on some tools (see above). If your rules.yaml matches
   on risk level, re-check those rules.

Also in this release

• docs/VERIFICATION.md — what the mock suite actually guarantees, a live
  verification checklist, and the criteria for claiming this tool verified.
• skills/ceph-aiops/references/agent-guardrails.md — for driving this tool with a
  smaller / local model: which guardrails are now enforced for you, and a
  ready-made system prompt for the rest.
• Expanded operator playbooks in the skill documentation.
• The advertised tool count now matches what an MCP client actually lists
  (it includes undo_list / undo_apply), and a release gate keeps it honest.
• The (preview) label has been dropped. It never meant unreleased; verification
  status now lives in docs/VERIFICATION.md where it can be specific.

v0.3.0

Undo executor (undo list / undo apply) line-wide + this tool's platform/framework expansion + coverage. See CHANGELOG.md.

v0.2.1

Fix release: secrets.enc follows _AIOPS_HOME; sanitized failures audit as status=error (no undo recorded for failed calls); doctor/init fully test-covered. Tool-specific fixes in CHANGELOG.md.

v0.2.0

Security-hardening release from a line-wide code review. Highlights: secure-by-default approver gate for high/critical operations (init seeds a starter rules.yaml), percent-encoded URL path segments, single-sourced version, governance persistence + CLI write-path tests. See CHANGELOG.md for the full list. BREAKING: fresh installs with no rules.yaml now deny high/critical writes until an approver is recorded.