memgit 0.5.0__tar.gz → 0.6.0__tar.gz

{memgit-0.5.0 → memgit-0.6.0}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: memgit
-Version: 0.5.0
+Version: 0.6.0
 Summary: Git for AI memory — version-controlled context persistence across Claude, GPT, Gemini, Cursor, Windsurf, and more
 License: MIT
 Project-URL: Homepage, https://memgit.dev
@@ -44,7 +44,7 @@ Version-controlled, cross-AI context that persists, diffs, rolls back, and syncs
 
 [![PyPI](https://img.shields.io/pypi/v/memgit)](https://pypi.org/project/memgit/)
 [![License: MIT](https://img.shields.io/badge/License-MIT-green.svg)](LICENSE)
-[![Tests](https://img.shields.io/badge/tests-160%20passing-brightgreen)](tests/)
+[![Tests](https://img.shields.io/badge/tests-245%20passing-brightgreen)](tests/)
 
 ---
 
@@ -215,13 +215,14 @@ memgit resume --json     # for tooling
 Wire it into Claude Code so memory becomes **automatic** — no tool call, no judgment required:
 
 ```bash
-memgit setup hooks       # installs all four hooks (~/.claude/settings.json)
+memgit setup hooks       # installs all five hooks (~/.claude/settings.json)
 ```
 
 | Hook | What it enforces |
 |---|---|
-| `SessionStart` | every session opens with the resume digest in context |
-| `UserPromptSubmit` | each prompt is BM25-matched against the store; relevant memories are injected (silent when nothing clears the relevance bar; never repeats within a session) — `--no-recall` to skip |
+| `SessionStart` | every session opens with the resume digest in context — status board, checkpoints, critical rules, memory index |
+| `UserPromptSubmit` | each prompt is BM25-matched against the store; relevant memories are injected, ending with a "+N more on '<topic>'" depth hint when more exists (silent when nothing clears the relevance bar; never repeats within a session) — `--no-recall` to skip |
+| `PostToolUse` | reading a file whose path matches a memory tag surfaces a one-line hint ("6 memories tagged 'x' relate to this path") — tagmap cache only, capped 3/session, `--no-ctx-recall` to skip |
 | `Stop` (guard) | a session that did real work but saved nothing gets ONE nudge to save durable facts before finishing — `--no-guard` to skip |
 | `Stop` (sync) | markdown memories are checkpointed asynchronously at session end |
 
@@ -294,13 +295,25 @@ And it **learns**: a sidecar usage ledger tracks which memories actually get rec
 
 ---
 
+## Depth advertisement, trackers & supersession (v0.6.0)
+
+Measured across 289 real sessions: injected recall reached ~59% of them, but only **6.8%** ever ran an active search — the injected top-3 reads as "memory consulted", so the model never learns there's a queryable store behind it. 0.6.0 makes the passive layer advertise what the active layer knows:
+
+- **Memory index** — the resume digest ends with tag→count pairs (`8a8f4ec (6) · instagram (5)`) and the exact call to go deeper. Counts are truthful: superseded memories are excluded, and every advertised topic is guaranteed to return search results.
+- **"+N more" recall hints** — when the per-prompt recall block has more on-topic memories behind it, it says so, with the one call to get them.
+- **Context-triggered recall** — a `PostToolUse` hook: reading a file whose path matches a memory tag surfaces `memgit: 6 memories tagged 'x' relate to this path`. Reads only a commit-time tagmap cache (never the store), capped 3/session.
+- **Trackers (`tr`)** — one memory per in-flight entity (`<entity>-status`), updated by re-saving the same slug. They render as a **status board** at the top of every session: memgit is the authority for entity status; files may lag.
+- **Supersession** — a correction names what it replaces (`supersedes=[old-slug]`) instead of a "CORRECTED:" prefix. Superseded memories vanish from search/recall/resume (history preserved; `list` still shows them marked ⊘), so injected context is never stale.
+
+---
+
 ## Commands
 
 ```bash
 # Core (git-like)
 memgit init                       # initialize store (auto-detects best path)
 memgit onboard                    # bootstrap brief for an existing codebase
-memgit add <slug> <rule>          # stage a memory (--body for full detail, --project to scope)
+memgit add <slug> <rule>          # stage a memory (--body detail, --project scope, --supersedes old-slug)
 memgit commit -m "message"        # checkpoint current state
 memgit log                        # history
 memgit diff [sha1] [sha2]         # what changed
@@ -432,7 +445,7 @@ git clone https://github.com/code4161/memgit.git
 cd memgit
 python -m venv .venv && source .venv/bin/activate
 pip install -e ".[dev]"
-pytest    # 160 tests, all passing, < 3 seconds
+pytest    # 245 tests, all passing, < 5 seconds
 ```
 
 See [CONTRIBUTING.md](CONTRIBUTING.md).

{memgit-0.5.0 → memgit-0.6.0}/README.md

@@ -10,7 +10,7 @@ Version-controlled, cross-AI context that persists, diffs, rolls back, and syncs
 
 [![PyPI](https://img.shields.io/pypi/v/memgit)](https://pypi.org/project/memgit/)
 [![License: MIT](https://img.shields.io/badge/License-MIT-green.svg)](LICENSE)
-[![Tests](https://img.shields.io/badge/tests-160%20passing-brightgreen)](tests/)
+[![Tests](https://img.shields.io/badge/tests-245%20passing-brightgreen)](tests/)
 
 ---
 
@@ -181,13 +181,14 @@ memgit resume --json     # for tooling
 Wire it into Claude Code so memory becomes **automatic** — no tool call, no judgment required:
 
 ```bash
-memgit setup hooks       # installs all four hooks (~/.claude/settings.json)
+memgit setup hooks       # installs all five hooks (~/.claude/settings.json)
 ```
 
 | Hook | What it enforces |
 |---|---|
-| `SessionStart` | every session opens with the resume digest in context |
-| `UserPromptSubmit` | each prompt is BM25-matched against the store; relevant memories are injected (silent when nothing clears the relevance bar; never repeats within a session) — `--no-recall` to skip |
+| `SessionStart` | every session opens with the resume digest in context — status board, checkpoints, critical rules, memory index |
+| `UserPromptSubmit` | each prompt is BM25-matched against the store; relevant memories are injected, ending with a "+N more on '<topic>'" depth hint when more exists (silent when nothing clears the relevance bar; never repeats within a session) — `--no-recall` to skip |
+| `PostToolUse` | reading a file whose path matches a memory tag surfaces a one-line hint ("6 memories tagged 'x' relate to this path") — tagmap cache only, capped 3/session, `--no-ctx-recall` to skip |
 | `Stop` (guard) | a session that did real work but saved nothing gets ONE nudge to save durable facts before finishing — `--no-guard` to skip |
 | `Stop` (sync) | markdown memories are checkpointed asynchronously at session end |
 
@@ -260,13 +261,25 @@ And it **learns**: a sidecar usage ledger tracks which memories actually get rec
 
 ---
 
+## Depth advertisement, trackers & supersession (v0.6.0)
+
+Measured across 289 real sessions: injected recall reached ~59% of them, but only **6.8%** ever ran an active search — the injected top-3 reads as "memory consulted", so the model never learns there's a queryable store behind it. 0.6.0 makes the passive layer advertise what the active layer knows:
+
+- **Memory index** — the resume digest ends with tag→count pairs (`8a8f4ec (6) · instagram (5)`) and the exact call to go deeper. Counts are truthful: superseded memories are excluded, and every advertised topic is guaranteed to return search results.
+- **"+N more" recall hints** — when the per-prompt recall block has more on-topic memories behind it, it says so, with the one call to get them.
+- **Context-triggered recall** — a `PostToolUse` hook: reading a file whose path matches a memory tag surfaces `memgit: 6 memories tagged 'x' relate to this path`. Reads only a commit-time tagmap cache (never the store), capped 3/session.
+- **Trackers (`tr`)** — one memory per in-flight entity (`<entity>-status`), updated by re-saving the same slug. They render as a **status board** at the top of every session: memgit is the authority for entity status; files may lag.
+- **Supersession** — a correction names what it replaces (`supersedes=[old-slug]`) instead of a "CORRECTED:" prefix. Superseded memories vanish from search/recall/resume (history preserved; `list` still shows them marked ⊘), so injected context is never stale.
+
+---
+
 ## Commands
 
 ```bash
 # Core (git-like)
 memgit init                       # initialize store (auto-detects best path)
 memgit onboard                    # bootstrap brief for an existing codebase
-memgit add <slug> <rule>          # stage a memory (--body for full detail, --project to scope)
+memgit add <slug> <rule>          # stage a memory (--body detail, --project scope, --supersedes old-slug)
 memgit commit -m "message"        # checkpoint current state
 memgit log                        # history
 memgit diff [sha1] [sha2]         # what changed
@@ -398,7 +411,7 @@ git clone https://github.com/code4161/memgit.git
 cd memgit
 python -m venv .venv && source .venv/bin/activate
 pip install -e ".[dev]"
-pytest    # 160 tests, all passing, < 3 seconds
+pytest    # 245 tests, all passing, < 5 seconds
 ```
 
 See [CONTRIBUTING.md](CONTRIBUTING.md).

{memgit-0.5.0 → memgit-0.6.0}/memgit/__init__.py

@@ -1,3 +1,3 @@
 """memgit — git for AI memory."""
 
-__version__ = "0.5.0"
+__version__ = "0.6.0"

{memgit-0.5.0 → memgit-0.6.0}/memgit/cli.py

@@ -152,9 +152,10 @@ def init(directory):
 @click.argument('slug')
 @click.argument('rule')
 @click.option('--type', '-t', 'type_code', default='fb',
-              type=click.Choice(['fb', 'us', 'pj', 'rf', 'cn', 'lx', 'co']),
+              type=click.Choice(['fb', 'us', 'pj', 'rf', 'cn', 'lx', 'co', 'tr']),
               help='fb=feedback us=user pj=project rf=reference cn=convention '
-                   'lx=lesson co=core (prefer `memgit core set` for co)')
+                   'lx=lesson co=core (prefer `memgit core set` for co) '
+                   'tr=tracker (live status of one entity; slug <entity>-status)')
 @click.option('--why', '-w', default=None, help='Reasoning / why this rule exists')
 @click.option('--when', '-W', default=None, help='When / where to apply')
 @click.option('--tags', default=None, help='Comma-separated tags')
@@ -165,7 +166,14 @@ def init(directory):
 @click.option('--project', '-P', default=None,
               help='Project this memory belongs to (default: derived from the '
                    'current directory; pass "" for a global memory)')
-def add(slug, rule, type_code, why, when, tags, priority, body, project):
+@click.option('--supersedes', default=None,
+              help='Comma-separated slugs this memory REPLACES (corrections, '
+                   'updated decisions) — superseded memories stop surfacing in '
+                   'search/recall/resume')
+@click.option('--related', default=None,
+              help='Comma-separated slugs of related memories')
+def add(slug, rule, type_code, why, when, tags, priority, body, project,
+        supersedes, related):
     """Add or update a mnemonic.
 
     SLUG  kebab-case identifier (e.g. ig-pipeline-no-fallback)\n
@@ -183,6 +191,12 @@ def add(slug, rule, type_code, why, when, tags, priority, body, project):
     elif not project.strip():
         project = None
 
+    from .links import validate_relations
+    sup_list, rel_list, warnings = validate_relations(
+        slug, supersedes, related, repo.list())
+    for w in warnings:
+        err.print(f'[yellow]{w}[/yellow]')
+
     m = Mnemonic(
         type_code=type_code,
         slug=slug,
@@ -194,10 +208,13 @@ def add(slug, rule, type_code, why, when, tags, priority, body, project):
         priority=priority,
         body=body,
         project=project,
+        supersedes=sup_list,
+        related=rel_list,
     )
     sha = repo.add(m)
     from rich.markup import escape as _mesc
-    console.print(f'[green]staged[/green]  {_mesc(m.slug)} {_mesc("[" + sha[:8] + "]")}')
+    sup_note = f'  supersedes {len(sup_list)}' if sup_list else ''
+    console.print(f'[green]staged[/green]  {_mesc(m.slug)} {_mesc("[" + sha[:8] + "]")}{sup_note}')
 
 
 # ── remove ────────────────────────────────────────────────────────────────────
@@ -540,6 +557,14 @@ def resume(checkpoints, recent, plain, fmt_json, project):
             console.print(f'  [red]removed[/red]  {s}')
         console.print()
 
+    if ctx.get('tracker_memories'):
+        console.print('[bold]Status board (live entity state):[/bold]')
+        for m in ctx['tracker_memories']:
+            ts = m['timestamp'].strftime('%m-%d')
+            rule = m['rule'][:80] + '..' if len(m['rule']) > 80 else m['rule']
+            console.print(f'  [red]●[/red] [cyan]{m["slug"]}[/cyan] [dim](upd {ts})[/dim] {rule}')
+        console.print()
+
     console.print('[bold]Last checkpoints:[/bold]')
     for ck in ctx['checkpoints']:
         ts = ck['timestamp'].strftime('%Y-%m-%d %H:%M')
@@ -557,6 +582,10 @@ def resume(checkpoints, recent, plain, fmt_json, project):
         console.print('\n[bold red]Critical rules (always apply):[/bold red]')
         for m in ctx['critical_memories']:
             console.print(f'  [red]![/red] [cyan]{m["slug"]}[/cyan]  {m["rule"]}')
+    if ctx.get('entity_index'):
+        idx = ' · '.join(f'{tag} ({n})' for tag, n in ctx['entity_index'])
+        console.print(f'\n[bold]Memory index:[/bold] {idx}  '
+                      f'[dim](memgit search "<topic>")[/dim]')
     if ctx.get('maintenance'):
         console.print(f'\n[yellow]maintenance:[/yellow] {ctx["maintenance"]}')
     console.print()
@@ -589,6 +618,19 @@ def _format_resume_plain(ctx: dict) -> str:
             body = (m.get('body') or m['rule']).rstrip()
             lines.append('')
             lines.append(body)
+    # Status board: live entity state. Rendered right after the guide because
+    # orientation ("what is the CURRENT state of things") is worth more than
+    # history. The freshness stamp is what tells the model to trust or
+    # re-verify a line.
+    if ctx.get('tracker_memories'):
+        lines.append('')
+        lines.append('## Status board — live entity state '
+                     '(memgit is authoritative; files may lag)')
+        for m in ctx['tracker_memories']:
+            ts = m['timestamp'].strftime('%m-%d')
+            lines.append(f'- {m["slug"]} (upd {ts}): {clip(m["rule"], 160)}')
+        lines.append('(State changed? Update the tracker: save_memory with the '
+                     'same slug.)')
     st = ctx['staged']
     if st['new'] or st['updated'] or st['removed']:
         lines.append('')
@@ -634,9 +676,26 @@ def _format_resume_plain(ctx: dict) -> str:
             'digest + seeding brief, then save 10-20 durable facts '
             '(purpose, architecture, conventions, state, gotchas) via save_memory.'
         )
+    if ctx.get('core_missing'):
+        lines.append('')
+        lines.append(
+            '(No core operating guide for this project yet — seed it once: '
+            '`memgit core seed`, review with `memgit core show`, then '
+            '`memgit core sync` to deliver it to every AI tool.)'
+        )
+    # Entity index LAST: recency position is where a model decides its next
+    # action, and this section exists purely to convert reading into querying.
+    if ctx.get('entity_index'):
+        lines.append('')
+        lines.append('## Memory index — depth beyond this digest')
+        lines.append('- ' + ' · '.join(f'{tag} ({n})'
+                                       for tag, n in ctx['entity_index']))
+        lines.append('(each is one call away: search_memories("<topic>", top_k=10))')
     lines.append('')
-    lines.append('(Check work-in-flight and the last checkpoints before assuming state; '
-                 'use memgit search for anything task-specific.)')
+    lines.append('(This digest is a teaser, not the memory. Trust the status board '
+                 'over files for entity state; check work-in-flight and the last '
+                 'checkpoints before assuming state. Anything deeper: '
+                 'search_memories("<topic from the index above>").)')
     return '\n'.join(lines)
 
 
@@ -661,6 +720,13 @@ def hook_stop_guard():
     sys.exit(stop_guard())
 
 
+@hook.command('context-recall')
+def hook_context_recall():
+    """PostToolUse: hint when memories exist about the file being read (stdin JSON)."""
+    from .hooks import context_recall
+    sys.exit(context_recall())
+
+
 # ── onboard ───────────────────────────────────────────────────────────────────
 
 ONBOARD_BRIEF = """\
@@ -683,6 +749,9 @@ Extract 10–20 DURABLE facts about this project and save each one as a memory
 - `rf` reference: key entry points, dashboards, external services, URLs
 - `fb` feedback: known constraints ("never touch X", "Y is production")
 - `lx` lesson: past incidents or gotchas documented in the repo
+- `tr` tracker: LIVE status of each in-flight entity (a deploy, a migration,
+  a draft) — slug `<entity>-status`; update it by re-saving the same slug.
+  memgit is the authority for this status; files may lag.
 
 Rules for good memories: one fact per memory; kebab-case slug; a one-line
 `rule` stating the fact; details in `body`; set `project` to "{project}";
@@ -936,21 +1005,31 @@ def show(slug, toon, fmt_markdown):
             console.print(f'[dim]Related: {_mesc(", ".join(m.related))}[/dim]')
         if m.supersedes:
             console.print(f'[dim]Supersedes: {_mesc(", ".join(m.supersedes))}[/dim]')
+        from .links import superseded_by, resolve_head
+        all_mems = repo.list()
+        heirs = superseded_by(m.slug, all_mems)
+        if heirs:
+            head = resolve_head(m.slug, all_mems)
+            console.print(f'[yellow]⊘ SUPERSEDED by: {_mesc(", ".join(heirs))}'
+                          f' — current head: {_mesc(head)}[/yellow]')
 
 
 # ── list ──────────────────────────────────────────────────────────────────────
 
 @cli.command(name='list')
 @click.option('--type', '-t', 'type_filter', default=None,
-              type=click.Choice(['fb', 'us', 'pj', 'rf', 'cn', 'lx']),
+              type=click.Choice(['fb', 'us', 'pj', 'rf', 'cn', 'lx', 'co', 'tr']),
               help='Filter by type')
 @click.option('--priority', '-p', default=None, type=click.IntRange(1, 3), help='Filter by priority')
 @click.option('--project', '-P', 'project_filter', default=None, help='Filter by project')
 @click.option('--toon', is_flag=True, help='Show TOON format')
 def list_cmd(type_filter, priority, project_filter, toon):
-    """List all mnemonics in the current thread."""
+    """List all mnemonics in the current thread (the audit view — superseded
+    memories are shown, marked ⊘)."""
+    from .links import superseded_slugs
     repo = _require_repo()
     mnemonics = repo.list()
+    hidden = superseded_slugs(mnemonics)
     if type_filter:
         mnemonics = [m for m in mnemonics if m.type_code == type_filter]
     if priority:
@@ -980,10 +1059,14 @@ def list_cmd(type_filter, priority, project_filter, toon):
         p_str = '!' if m.priority == 3 else str(m.priority)
         proj = (m.project or '')[:18]
         rule_preview = m.rule[:58] + '..' if len(m.rule) > 58 else m.rule
+        if m.slug in hidden:
+            rule_preview = '⊘ ' + rule_preview
         table.add_row(m.slug, m.type_code, p_str, proj, rule_preview)
 
     console.print(table)
-    console.print(f'\n[dim]{len(mnemonics)} mnemonic{"s" if len(mnemonics) != 1 else ""}[/dim]')
+    shown_hidden = sum(1 for m in mnemonics if m.slug in hidden)
+    hid_note = f' · {shown_hidden} superseded (⊘)' if shown_hidden else ''
+    console.print(f'\n[dim]{len(mnemonics)} mnemonic{"s" if len(mnemonics) != 1 else ""}{hid_note}[/dim]')
 
 
 # ── import ────────────────────────────────────────────────────────────────────
@@ -1183,11 +1266,15 @@ import re  # noqa: E402 — needed for lint command
 @click.option('--toon', is_flag=True, help='Output TOON format (token-efficient)')
 @click.option('--json', 'fmt_json', is_flag=True, help='Output JSON')
 @click.option('--type', '-t', 'type_filter', default=None,
-              type=click.Choice(['fb', 'us', 'pj', 'rf', 'cn', 'lx']),
+              type=click.Choice(['fb', 'us', 'pj', 'rf', 'cn', 'lx', 'co', 'tr']),
               help='Filter by type before scoring')
 @click.option('--project', '-P', 'project_filter', default=None,
               help='Only memories from this project (as shown in `memgit list`)')
-def search(query, top, toon, fmt_json, type_filter, project_filter):
+@click.option('--include-superseded', is_flag=True,
+              help='Also score memories that a newer memory supersedes '
+                   '(hidden by default)')
+def search(query, top, toon, fmt_json, type_filter, project_filter,
+           include_superseded):
     """Search memories by relevance.
 
     Returns the top-k mnemonics scored against QUERY using BM25.
@@ -1197,6 +1284,9 @@ def search(query, top, toon, fmt_json, type_filter, project_filter):
 
     repo = _require_repo()
     mnemonics = repo.list()
+    if not include_superseded:
+        from .links import filter_active
+        mnemonics = filter_active(mnemonics)
     if type_filter:
         mnemonics = [m for m in mnemonics if m.type_code == type_filter]
     if project_filter:
@@ -1359,7 +1449,7 @@ def graph(output, auto_open):
     """Generate an interactive HTML graph of the memory store.
 
     Visualizes all mnemonics as a force-directed graph with:
-      - Nodes colored by type (fb/us/pj/rf/cn/lx)
+      - Nodes colored by type (fb/us/pj/rf/cn/lx/co/tr)
       - Node size by priority
       - Edges from [[wikilink]] references and explicit related/supersedes links
       - Filter by type, search by keyword, click to highlight neighbours
@@ -1468,7 +1558,8 @@ def stats(fmt_json):
         return
 
     type_labels = {'fb': 'feedback', 'us': 'user', 'pj': 'project',
-                   'rf': 'reference', 'cn': 'convention', 'lx': 'lesson'}
+                   'rf': 'reference', 'cn': 'convention', 'lx': 'lesson',
+                   'co': 'core', 'tr': 'tracker'}
 
     type_str = ' · '.join(
         f"{s['by_type'].get(tc, 0)} {lbl}"
@@ -2245,7 +2336,8 @@ def setup_gemini_cli(dry_run):
 
 
 #: substrings identifying a hook command as one of ours (any generation)
-_MEMGIT_HOOK_SIGNS = ('resume --plain', 'hook prompt-recall', 'hook stop-guard', ' sync')
+_MEMGIT_HOOK_SIGNS = ('resume --plain', 'hook prompt-recall', 'hook stop-guard',
+                      'hook context-recall', ' sync')
 
 
 def _is_memgit_hook_entry(h: dict) -> bool:
@@ -2262,19 +2354,25 @@ def _is_memgit_hook_entry(h: dict) -> bool:
               help='Skip the per-prompt auto-recall hook (UserPromptSubmit)')
 @click.option('--no-guard', is_flag=True,
               help='Skip the end-of-session capture guard (Stop)')
+@click.option('--no-ctx-recall', is_flag=True,
+              help='Skip the context-recall hook (PostToolUse on Read/Grep/Glob)')
 @click.option('--dry-run', is_flag=True, help='Show the change without writing')
-def setup_hooks(remove, no_recall, no_guard, dry_run):
+def setup_hooks(remove, no_recall, no_guard, no_ctx_recall, dry_run):
     """Install the Claude Code hooks that make memory automatic.
 
-    Four hooks, one principle: what a hook enforces happens, what a tool
+    Five hooks, one principle: what a hook enforces happens, what a tool
     description suggests mostly doesn't (measured: 6% voluntary engagement
     vs 100% hook delivery).
 
     \b
-      SessionStart      inject `memgit resume` — last checkpoints, work in
-                        flight, critical rules
+      SessionStart      inject `memgit resume` — status board, last
+                        checkpoints, work in flight, critical rules,
+                        memory index
       UserPromptSubmit  inject memories relevant to each prompt (BM25,
                         silent when nothing clears the relevance bar)
+      PostToolUse       context recall — reading a file whose path matches
+                        a memory tag surfaces a one-line depth hint
+                        (tagmap cache only, capped 3/session)
       Stop              capture guard — a substantive session ending with
                         zero memory writes gets ONE nudge to save durable
                         facts; plus async `memgit sync` to checkpoint
@@ -2313,6 +2411,10 @@ def setup_hooks(remove, no_recall, no_guard, dry_run):
             {'type': 'command',
              'command': f'{base} hook prompt-recall 2>/dev/null || true'},
         ],
+        'PostToolUse': [] if no_ctx_recall else [
+            {'type': 'command',
+             'command': f'{base} hook context-recall 2>/dev/null || true'},
+        ],
         'Stop': ([] if no_guard else [
             {'type': 'command',
              'command': f'{base} hook stop-guard 2>/dev/null || true'},
@@ -2322,14 +2424,20 @@ def setup_hooks(remove, no_recall, no_guard, dry_run):
              'async': True},
         ],
     }
+    #: PostToolUse fires per tool call — matcher limits it to the file-reading
+    #: tools so the hook binary isn't spawned on every Bash/Edit.
+    _MATCHERS = {'PostToolUse': 'Read|Grep|Glob'}
 
     changed = []
-    for event in ('SessionStart', 'UserPromptSubmit', 'Stop'):
+    for event in ('SessionStart', 'UserPromptSubmit', 'PostToolUse', 'Stop'):
         entries = hooks.setdefault(event, [])
         had = [h for h in entries if isinstance(h, dict) and _is_memgit_hook_entry(h)]
         entries[:] = [h for h in entries if h not in had]
         if not remove and plan[event]:
-            entries.append({'hooks': plan[event]})
+            entry: dict = {'hooks': plan[event]}
+            if event in _MATCHERS:
+                entry['matcher'] = _MATCHERS[event]
+            entries.append(entry)
             changed.append(event)
         if not entries:
             hooks.pop(event, None)
@@ -2349,7 +2457,7 @@ def setup_hooks(remove, no_recall, no_guard, dry_run):
             tag = ' [dim](async)[/dim]' if inner.get('async') else ''
             console.print(f'  [cyan]{event}[/cyan]  [dim]{inner["command"]}[/dim]{tag}')
     console.print('[dim]Resume at session start, relevant memories per prompt, '
-                  'capture guard + sync at stop.[/dim]')
+                  'context hints on file reads, capture guard + sync at stop.[/dim]')
 
 
 @setup.command('print-config')

{memgit-0.5.0 → memgit-0.6.0}/memgit/delivery.py

@@ -122,13 +122,29 @@ def _upsert_marker_block(existing: str, block_body: str) -> str:
 # ── seed: ingest existing host skills/rules into a draft core guide ───────────
 
 def _frontmatter_field(text: str, key: str) -> Optional[str]:
-    """Pull a top-level `key: value` from a leading `---` YAML frontmatter."""
+    """Pull a top-level `key: value` from a leading `---` YAML frontmatter.
+
+    Handles folded/literal block scalars (`key: >` / `key: |`): the value is
+    the first indented line that follows — one line is enough for a routing
+    entry, and a real YAML parser is not worth the dependency here.
+    """
     if not text.startswith("---"):
         return None
     end = text.find("\n---", 3)
     block = text[3:end] if end != -1 else text[3:]
-    m = re.search(rf"^{re.escape(key)}:\s*(.+)$", block, flags=re.M)
-    return m.group(1).strip().strip('"\'') if m else None
+    m = re.search(rf"^{re.escape(key)}:\s*(.*)$", block, flags=re.M)
+    if not m:
+        return None
+    value = m.group(1).strip().strip('"\'')
+    if value in (">", "|", ">-", "|-", ""):
+        tail = block[m.end():]
+        for line in tail.splitlines():
+            if line.strip() and line[:1] in (" ", "\t"):
+                return line.strip()
+            if line.strip():  # next top-level key — no scalar body
+                break
+        return None
+    return value
 
 
 #: (label, glob) pairs — where each host keeps model-invoked skills.
@@ -173,7 +189,13 @@ def build_seed(root: Path, home: Optional[Path] = None) -> str:
         "## memgit",
         "- Before answering anything that depends on past work, call resume/search "
         "(memgit) — the record of prior sessions lives there, not in this file.",
-        "- Save durable facts/decisions/lessons as you learn them.",
+        "- memgit is the authority for entity STATUS ('tr' tracker memories: "
+        "deploys, drafts, migrations, campaigns); files and READMEs are "
+        "downstream and may lag. Changed an entity's state? Update its "
+        "<entity>-status tracker (save, same slug).",
+        "- Save durable facts/decisions/lessons as you learn them. A memory "
+        "that corrects an old one should supersede it (supersedes=[old-slug]), "
+        "not sit beside it.",
     ]
 
     skills = collect_skills(root, home)
@@ -235,11 +257,18 @@ def compute_auto_section(repo, project, now, curated: str = "") -> str:
     if not usage:
         return ""
     candidates = []
-    for m in repo.list():
+    from .links import superseded_slugs
+    all_mems = repo.list()
+    hidden = superseded_slugs(all_mems)
+    for m in all_mems:
         if m.type_code in ("co", "cn"):        # skip core + conventions (rules)
             continue
+        if m.type_code == "tr":                # skip trackers — live state must
+            continue                           # never fossilize in static files
         if m.priority == 3:                    # skip always-on criticals (rules)
             continue
+        if m.slug in hidden:                   # skip superseded (stale by definition)
+            continue
         if project and project_affinity(m.project, project) < 1:
             continue
         if not project and m.project:

{memgit-0.5.0 → memgit-0.6.0}/memgit/graph.py

@@ -17,6 +17,8 @@ _TYPE_COLOR = {
     "rf": "#8b5cf6",  # violet   — reference
     "cn": "#eab308",  # yellow   — convention
     "lx": "#ec4899",  # pink     — lesson
+    "co": "#64748b",  # slate    — core guide
+    "tr": "#ef4444",  # red      — tracker (live status)
 }
 _TYPE_LABEL = {
     "fb": "feedback",
@@ -25,6 +27,8 @@ _TYPE_LABEL = {
     "rf": "reference",
     "cn": "convention",
     "lx": "lesson",
+    "co": "core",
+    "tr": "tracker",
 }
 
 _WIKILINK_RE = re.compile(r'\[\[([a-z0-9_-]+)\]\]', re.IGNORECASE)
@@ -372,7 +376,7 @@ window.addEventListener('resize', () => {
 const tooltip = document.getElementById('tooltip');
 
 node.on('mouseover', (event, d) => {
-  const tags = d.tags.filter(t => !['fb','pj','us','rf','cn','lx'].includes(t));
+  const tags = d.tags.filter(t => !['fb','pj','us','rf','cn','lx','co','tr'].includes(t));
   tooltip.innerHTML = `
     <div class="t-slug">${d.id}</div>
     <div class="t-type">[${d.type}] ${d.type_label} · priority ${d.priority}</div>