Integrate csb with dz claude-session-metadata catalog
Context
Recuva recovery of the 654 missing jsonls (Jan 23 - Apr 2 gap) failed -- SSD TRIM wiped deleted sectors. However, dz claude-session-metadata (formerly claude-lost-sessions) has built a structured catalog at ~/claude/lost-sessions/ with 156 lost session entries. Each entry contains:
summary.md -- UUID, start/end times, duration, command counts, tools used, first/last commands
known-docs/ -- symlinks to authored/edited markdown docs from that session's timeframe
folders-worked-on/ -- symlinks to project directories the session touched
sesslog@ -- symlink to the original sesslog folder
This data is recoverable from sesslogs + private/claude/ folders even when the transcript.jsonl is gone. See design analysis at 2026-04-08__07-41-13__full-context-sessions-and-csb-integration.md in the dazzlecmd project.
Brainstorms / Quick Todos
1. csb show <id> enrichment
When showing a session, check if ~/claude/lost-sessions/<timestamp>__*__<uuid-prefix>/ exists. If so, display:
[LOST] marker at top
- Known-docs count + clickable paths
- Folders-worked-on count
- Link to the catalog summary.md
For live sessions, check ~/claude/live-sessions-test/ (or wherever the future full-sessions/ catalog lives).
2. csb list lost/live markers
Add [LOST] / [LIVE] flags next to session names when the catalog has an entry. Could be a column or inline tag:
1. CHROME-PLUGIN__jump-to-my-reply [LIVE+CATALOG] today (2026-04-10)
2. cc-hooks-names-and-logging [LOST+CATALOG] 2026-01-24
3. (some session) [LIVE] (no catalog entry yet)
3. csb catalog <id> -- explicit catalog view
New subcommand that renders the catalog summary.md with rich formatting. Could open in browser or display in terminal with rich markdown.
4. csb catalog --build -- trigger metadata tool
Wrapper that invokes dz claude-session-metadata --all --apply from csb. Convenient if a user doesn't have dz on PATH or doesn't know about the separate tool.
5. Catalog path configuration
Add config keys for catalog paths:
{
"catalog_paths": {
"lost": "~/claude/lost-sessions",
"live": "~/claude/full-sessions"
}
}
This way the integration is loosely coupled -- csb reads from wherever the catalog lives.
6. csb resume <id> catalog awareness
Before resuming, check if the session has a catalog entry with known-docs. If so, print them so the user has context:
Resuming: cc-hooks-names-and-logging
Known docs from this session:
- 2026-01-29__03-27-06__full-postmortem_claude-session-logger-initial-release.md
- 2026-01-29__14-46-10__bidirectional-project-sesslog-linking.md
Folders worked on:
- C:\code\claude-projects\claude-session-logger\github
- C:\code\claude-projects\claude-session-logger\local
claude --resume 833a100e-d959-47aa-9db2-d22fdb6d7659
7. SQLite index of catalog data
For fast csb search across known-docs/folders, we could index the catalog data into the csb SQLite DB. New tables:
catalog_entries (session_id, catalog_path, has_transcript, built_at)
known_docs (session_id, doc_path, exists)
catalog_folders (session_id, folder_path, exists)
Then csb search "plugin architecture" could hit both session names AND authored docs from those sessions.
8. Archaeological view -- cross-session doc traversal
Given that known-docs symlinks exist, we can trace "which sessions created/edited this doc?":
csb doc-sessions "2026-01-29__03-27-06__full-postmortem_claude-session-logger-initial-release.md"
> cc-hooks-names-and-logging (833a100e) [LOST] -- created
> (some other session) (47961256) [LIVE] -- edited
Priority ordering (rough)
| # |
Feature |
Effort |
Value |
| 5 |
Config for catalog paths |
Low |
Foundation for all others |
| 1 |
csb show enrichment |
Low |
High (immediate visibility) |
| 2 |
csb list [LOST]/[LIVE] markers |
Low |
High (at-a-glance status) |
| 6 |
csb resume context print |
Low |
Medium (resume UX) |
| 3 |
csb catalog <id> |
Medium |
Medium |
| 4 |
csb catalog --build wrapper |
Low |
Low (dz already works) |
| 7 |
SQLite catalog index |
High |
High (enables fast search) |
| 8 |
csb doc-sessions |
Medium |
Niche but powerful |
Prerequisites
Related
Integrate csb with dz claude-session-metadata catalog
Context
Recuva recovery of the 654 missing jsonls (Jan 23 - Apr 2 gap) failed -- SSD TRIM wiped deleted sectors. However,
dz claude-session-metadata(formerly claude-lost-sessions) has built a structured catalog at~/claude/lost-sessions/with 156 lost session entries. Each entry contains:summary.md-- UUID, start/end times, duration, command counts, tools used, first/last commandsknown-docs/-- symlinks to authored/edited markdown docs from that session's timeframefolders-worked-on/-- symlinks to project directories the session touchedsesslog@-- symlink to the original sesslog folderThis data is recoverable from sesslogs +
private/claude/folders even when the transcript.jsonl is gone. See design analysis at2026-04-08__07-41-13__full-context-sessions-and-csb-integration.mdin the dazzlecmd project.Brainstorms / Quick Todos
1.
csb show <id>enrichmentWhen showing a session, check if
~/claude/lost-sessions/<timestamp>__*__<uuid-prefix>/exists. If so, display:[LOST]marker at topFor live sessions, check
~/claude/live-sessions-test/(or wherever the futurefull-sessions/catalog lives).2.
csb listlost/live markersAdd
[LOST]/[LIVE]flags next to session names when the catalog has an entry. Could be a column or inline tag:3.
csb catalog <id>-- explicit catalog viewNew subcommand that renders the catalog summary.md with rich formatting. Could open in browser or display in terminal with rich markdown.
4.
csb catalog --build-- trigger metadata toolWrapper that invokes
dz claude-session-metadata --all --applyfrom csb. Convenient if a user doesn't have dz on PATH or doesn't know about the separate tool.5. Catalog path configuration
Add config keys for catalog paths:
{ "catalog_paths": { "lost": "~/claude/lost-sessions", "live": "~/claude/full-sessions" } }This way the integration is loosely coupled -- csb reads from wherever the catalog lives.
6.
csb resume <id>catalog awarenessBefore resuming, check if the session has a catalog entry with known-docs. If so, print them so the user has context:
7. SQLite index of catalog data
For fast
csb searchacross known-docs/folders, we could index the catalog data into the csb SQLite DB. New tables:catalog_entries(session_id, catalog_path, has_transcript, built_at)known_docs(session_id, doc_path, exists)catalog_folders(session_id, folder_path, exists)Then
csb search "plugin architecture"could hit both session names AND authored docs from those sessions.8. Archaeological view -- cross-session doc traversal
Given that known-docs symlinks exist, we can trace "which sessions created/edited this doc?":
Priority ordering (rough)
csb showenrichmentcsb list[LOST]/[LIVE] markerscsb resumecontext printcsb catalog <id>csb catalog --buildwrappercsb doc-sessionsPrerequisites
dz claude-session-metadata --allflag for unified lost+live catalog~/claude/sessions/unified? orlost-sessions/+full-sessions/?)YYYY-MM-DD__HH-MM__NAME__UUID-PREFIX/)Related
2026-04-08__07-41-13__full-context-sessions-and-csb-integration.md(in dazzlecmd)