docs: Gap analysis — claude-mem vs SerialMemory feature comparison#20
Merged
sblanchard merged 6 commits intomasterfrom Feb 22, 2026
Merged
Conversation
Comprehensive analysis of features in claude-mem that could improve SerialMemory's Claude Code integration. Identifies 10 gaps across auto-capture hooks, progressive disclosure, context injection, structured observations, privacy tags, timeline navigation, token tracking, plugin packaging, transcript integration, and web UI. Key P0 recommendations: token-efficient progressive search, automatic session capture via lifecycle hooks, and automatic context injection at session start. https://claude.ai/code/session_01M5GUmiM8WehUBUvKWiWgcJ
… structured observations, privacy tags P0 — Progressive Disclosure (GAP 2): - Add memory_search_index tool: compact index search (~50-80 tokens/result vs ~500+) - Add memory_timeline tool: chronological navigation around anchor point - Add memory_fetch tool: batch fetch full details by IDs - 3-layer workflow saves ~10x tokens: search_index → timeline → fetch - Token estimation on all search results P0 — Automatic Context Injection (GAP 3): - Update SessionStart hooks with progressive disclosure workflow guidance - Hook scripts updated with new tool matchers P0 — Automatic Session Capture (GAP 1): - Hook scripts updated for 15 tool matchers (was 12) - Progressive disclosure tools integrated into PreToolUse/PostToolUse hooks P1 — Structured Observations (GAP 5): - Add title, facts, concepts, files_read, files_modified to Memory model - Auto-generate title from first line when not provided - memory_ingest accepts structured fields for richer data capture - Database migration: ops/migrate_structured_observations.sql P1 — Token Budget Tracking (GAP 4): - EstimatedTokens field on MemorySearchResult - SearchMeta with IndexTokens, FullFetchTokens, TokenSavingsPercent - Compact search results show token savings percentage P2 — Privacy Tags (GAP 6): - <private>...</private> content automatically stripped before storage - Added to memory_ingest tool description Files changed: - New: ProgressiveDisclosureTools.cs, migrate_structured_observations.sql - Modified: Memory model, IKnowledgeGraphStore, PostgresKnowledgeGraphStore, KnowledgeGraphService, CoreToolHandlers, CoreToolDefinitions, ToolDefinitions, ToolHierarchy, Program.cs, install-hooks.sh https://claude.ai/code/session_01M5GUmiM8WehUBUvKWiWgcJ
Memories now auto-refresh every 10s (when showing recent, not search results) and new items get a cyan glow animation. Auto-scrolls to top unless the user has scrolled down manually. Shows a "+N new" badge when fresh memories arrive. https://claude.ai/code/session_01M5GUmiM8WehUBUvKWiWgcJ
Adds 10 new REST endpoints matching recently added MCP tools:
Progressive Disclosure:
- GET /api/memories/search-index (compact token-efficient search)
- GET /api/memories/timeline (chronological navigation around anchor)
- POST /api/memories/fetch (batch fetch full details by IDs)
Goal Management:
- POST /api/goals (set/update a goal)
- GET /api/goals (list active goals)
- POST /api/goals/{key}/complete (mark goal complete)
Summarization (requires LLM):
- POST /api/summarize/session (summarize session memories)
- POST /api/summarize/context (summarize recent context)
Auto-Capture:
- GET /api/captures/status (check capture buffer status)
- POST /api/captures/drain (drain JSONL entries into memories)
https://claude.ai/code/session_01M5GUmiM8WehUBUvKWiWgcJ
…system) Captures are now stored in a `session_captures` DB table instead of local JSONL files. All operations go through HTTP — no filesystem dependency on the server. New/updated endpoints: - POST /api/captures — batch-store capture entries - POST /api/captures/entry — store a single capture entry - GET /api/captures/status — buffer status from DB - POST /api/captures/drain — drain staged captures into memories Stack: - ops/migrate_session_captures.sql — table with RLS + indexes - ICaptureStore interface + PostgresKnowledgeGraphStore impl - KnowledgeGraphService.DrainCapturesAsync with tool-type grouping - SessionCapture model https://claude.ai/code/session_01M5GUmiM8WehUBUvKWiWgcJ
MCP submodule updated with 16 new tool routes + schemas for progressive disclosure, goals, server-side captures, and summarization. ops/session-capture.sh — now POSTs to /api/captures/entry over HTTP instead of writing local JSONL files. Uses curl with 2s timeout. Reads SERIALMEMORY_ENDPOINT and SERIALMEMORY_API_KEY from env. ops/session-summarize.sh — now reads capture status from /api/captures/status over HTTP instead of local filesystem. ops/install-hooks.sh — added PreToolUse/PostToolUse matchers for drain_session_captures, capture_status, goal_set/list/complete, summarize_session, summarize_context (23 total matchers). https://claude.ai/code/session_01M5GUmiM8WehUBUvKWiWgcJ
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Comprehensive analysis of features in claude-mem that could improve
SerialMemory's Claude Code integration. Identifies 10 gaps across
auto-capture hooks, progressive disclosure, context injection, structured
observations, privacy tags, timeline navigation, token tracking, plugin
packaging, transcript integration, and web UI.
Key P0 recommendations: token-efficient progressive search, automatic
session capture via lifecycle hooks, and automatic context injection
at session start.
https://claude.ai/code/session_01M5GUmiM8WehUBUvKWiWgcJ