Copilot review on #3 caught that the docstring claimed
the recovery collection was "now-deleted," but this PR only removes
the write call site — collection retirement is the follow-up PR #5.
Reword to match actual scope: the recovery-marker write is removed
here; the collection still exists and gets disposed in #5.

Co-Authored-By: Claude Opus 4.7 (1M context) <[email protected]>

* fix(hooks): drop checkpoint write path from Stop + PreCompact

The Stop hook used to do two things on each fire:
  (a) write a 1-KB checkpoint summary diary entry into the
      mempalace_session_recovery collection
  (b) auto-mine the verbatim transcript into mempalace_drawers

The summary in (a) is redundant once (b) is searchable, AND the
recovery collection had no semantic-search MCP surface — so
checkpoints were structurally invisible to mempalace_search. Net
effect from a user's seat: agents (and JP) couldn't find recent
session content via search even though everything was on disk.

Drop the summary half. Verbatim transcript chunks already contain
every word a checkpoint summary would have surfaced. Leaving only
the transcript-ingest path matches the verbatim-only architecture
described in docs/superpowers/specs/2026-05-05-verbatim-only-design.md.

Changes:

* hook_stop silent path: remove _save_diary_direct call. Save marker
  advances unconditionally (fire-and-forget mining doesn't have a
  "confirmed save" signal to gate on; failure detection moves to
  daemon-side observability — hook.log + systemd journal).
* hook_precompact: remove the recovery-marker write. Mine + compaction
  proceed as before.
* systemMessage shape changes from "✦ N memories woven into the
  palace — themes" to "✦ Transcript ingest triggered (wing=...)".
  Lighter, accurate, no false count.
* Delete the now-unused _save_diary_direct function (~120 LOC) and
  its dependencies _extract_themes + _THEME_STOPWORDS (~30 LOC).
* Update 4 hook tests + 1 OSError test to mock _ingest_transcript
  rather than _save_diary_direct, and to expect the new
  systemMessage shape.

Scope deliberately limited: this PR removes hook-side WRITES.
mempalace_session_recovery collection still exists; the
mempalace_session_recovery_read MCP tool still works against the
existing 763 archived entries; mempalace.migrate is untouched.
A follow-up PR will dispose of the collection itself once this
change has run in production for a verification window.

Net diff: −230 / +53 LOC. 1558 tests pass, 1 skipped, lint clean.

Co-Authored-By: Claude Opus 4.7 (1M context) <[email protected]>

* docs(spec): verbatim-only architecture design

Captures the architectural shift driving this PR family — drop
checkpoint summaries, expose only the verbatim transcripts that
mempalace_search can already reach. Documents:

* Why: the recovery collection had no semantic-search MCP surface,
  so checkpoints in it were structurally invisible to search
* What: hook write-path deletion (this PR) + collection disposal
  (follow-up PR)
* How: code-deletion checklist for both repos, migration plan with
  tarball safety net, test plan
* Open questions queued for review at merge time

Spec gates the destructive parts on the answers to those open
questions; the write-path removal in this PR is scoped narrow
enough to land without them.

Co-Authored-By: Claude Opus 4.7 (1M context) <[email protected]>

* docs(hooks): clarify hook_precompact docstring scope

Copilot review on #3 caught that the docstring claimed
the recovery collection was "now-deleted," but this PR only removes
the write call site — collection retirement is the follow-up PR #5.
Reword to match actual scope: the recovery-marker write is removed
here; the collection still exists and gets disposed in #5.

Co-Authored-By: Claude Opus 4.7 (1M context) <[email protected]>

---------

Co-authored-by: Claude Opus 4.7 (1M context) <[email protected]>

Follow-up to fix/drop-checkpoint-write-path. With nothing writing to
the recovery collection anymore (the hooks moved to verbatim-only on
the parent branch), the read paths and the migration that fed the
collection are dead code. Delete them.

Removed in mempalace/:
- palace.py: _SESSION_RECOVERY_COLLECTION constant,
  get_session_recovery_collection(), _CHECKPOINT_TOPICS tuple
- mcp_server.py: _get_session_recovery_collection(),
  _recovery_collection_cache global, the topic-routing branch in
  tool_diary_write (now always lands in mempalace_drawers),
  tool_session_recovery_read() and its TOOLS dict registration
- migrate.py: migrate_checkpoints_to_recovery() and the dependent
  _CHECKPOINT_TOPICS import
- cli.py: cmd_repair --mode reorganize handler + the choice flag

Removed in tests/:
- tests/test_session_recovery.py — entire file (recovery-collection
  module test suite)
- tests/test_migrate.py: TestMigrateCheckpointsToRecovery class
- tests/test_mcp_server.py: TestCheckpointRouting and
  TestSessionRecoveryRead classes

Removed in docs/:
- website/reference/mcp-tools.md: mempalace_session_recovery_read
  section (caught by the readme-claims meta-test before deploy)

Production data on disks still has the mempalace_session_recovery
collection with its 763 archived entries — this PR's code change
makes the collection unreachable through any MCP/CLI path, but does
NOT delete the on-disk data. A separate one-shot script
(scripts/phase2_purge_recovery.py per the spec) handles the
collection-level delete after deploy. JP signed off on hard-delete
in the spec ack.

Net diff: 12 files, −1300 lines / +30 lines. 1540 tests pass, 1
skipped. Ruff lint + format clean.

Stacked on #3 (which stops new writes); will
auto-rebase against main when that merges.

Co-Authored-By: Claude Opus 4.7 (1M context) <[email protected]>

…ol (#8)

* refactor(mcp): retire mempalace_session_recovery collection + read tool

Follow-up to fix/drop-checkpoint-write-path. With nothing writing to
the recovery collection anymore (the hooks moved to verbatim-only on
the parent branch), the read paths and the migration that fed the
collection are dead code. Delete them.

Removed in mempalace/:
- palace.py: _SESSION_RECOVERY_COLLECTION constant,
  get_session_recovery_collection(), _CHECKPOINT_TOPICS tuple
- mcp_server.py: _get_session_recovery_collection(),
  _recovery_collection_cache global, the topic-routing branch in
  tool_diary_write (now always lands in mempalace_drawers),
  tool_session_recovery_read() and its TOOLS dict registration
- migrate.py: migrate_checkpoints_to_recovery() and the dependent
  _CHECKPOINT_TOPICS import
- cli.py: cmd_repair --mode reorganize handler + the choice flag

Removed in tests/:
- tests/test_session_recovery.py — entire file (recovery-collection
  module test suite)
- tests/test_migrate.py: TestMigrateCheckpointsToRecovery class
- tests/test_mcp_server.py: TestCheckpointRouting and
  TestSessionRecoveryRead classes

Removed in docs/:
- website/reference/mcp-tools.md: mempalace_session_recovery_read
  section (caught by the readme-claims meta-test before deploy)

Production data on disks still has the mempalace_session_recovery
collection with its 763 archived entries — this PR's code change
makes the collection unreachable through any MCP/CLI path, but does
NOT delete the on-disk data. A separate one-shot script
(scripts/phase2_purge_recovery.py per the spec) handles the
collection-level delete after deploy. JP signed off on hard-delete
in the spec ack.

Net diff: 12 files, −1300 lines / +30 lines. 1540 tests pass, 1
skipped. Ruff lint + format clean.

Stacked on #3 (which stops new writes); will
auto-rebase against main when that merges.

Co-Authored-By: Claude Opus 4.7 (1M context) <[email protected]>

* docs(README): align with verbatim-only retirement

Copilot review on #5 caught that the README still
documented session_recovery_read as current, which would leave users
calling a tool that returns "Unknown tool" after this PR ships.

Updates the four user-facing sections that described the recovery
collection as a current shipping feature:

* "What just shipped" — adds 2026-05-05 update note explaining the
  split was retired, with the generalizable lesson preserved (a side
  collection without a semantic-search MCP read tool is invisible).
* "Verbatim vs derivative" principle — reframes the recovery
  collection as the failed first attempt rather than a current
  example; pattern stays valid but each future sibling needs its
  own read surface.
* "Structural retrieval fixes" bullet — describes verbatim-only
  end-state, links to the new spec, retains historical context for
  Apr 25 → May 5.
* "Deterministic hook saves" bullet — updates to the new ingest-only
  save path, removes "recovery-collection marker" mention.
* "What it looks like in production" code block — updates the
  systemMessage example to the new "Transcript ingest triggered
  (wing=...)" shape.
* "P8" architectural principle — keeps it but reframes as on-hold
  with a precondition (read-surface parity) drawn from the cycle.
* Fork-ahead tracking table — replaces the old multi-collection-split
  row with a verbatim-only retirement row pointing at the four PRs;
  also retires references to session_recovery_read in the drawer_id
  surfacing row; adds new row for the mining management CLI (#4).

42 README meta-tests pass (test_readme_claims.py).

Co-Authored-By: Claude Opus 4.7 (1M context) <[email protected]>

---------

Co-authored-by: Claude Opus 4.7 (1M context) <[email protected]>