$ timeahead.in
/ servers/github/carrymem
github

carrymem

Your portable AI memory — preferences, decisions, and corrections that follow you across models, tools, and devices. Zero-dependency, local-first, 85% PrefEval accuracy.

15critical
14 credentials detected in repository history via Gitleaks
▣ Score BreakdownMCPScore = Σ(raw × weight)
DimensionRawWeighted
Security
35%
0
0.0
Freshness
25%
30
7.5
Adoption
20%
0
0.0
Quality
10%
20
2.0
Trust
10%
50
5.0
Total
14.5
⚿ Capabilities & Risk Explainer
fs readfs writenetworkexecevalsecrets
◆ Risk level: high· 27 tools · auth: API key
fs read + fs write + network + exec + eval + secrets active — can execute code, access credentials, and make external network calls.
Tool nameDescriptionDestructive?
classify_messageAnalyze a user message and classify it into one of 7 memory types: user_preference (⭐), correction (🔧), decision (🎯), ✓ no
get_classification_schemaReturn CarryMem's complete classification schema: all 7 memory types with their icons, descriptions, examples, confidenc✓ no
batch_classifyClassify multiple messages in a single call. Accepts an array of messages and returns classifications for each. More eff✓ no
classify_and_rememberClassify a user message into one of 7 memory types AND store it if it contains memorable information (should_remember=tr✓ no
recall_memoriesSearch stored memories using semantic full-text search (FTS5) across all memory types. Supports filtering by namespace, ✓ no
+22 more tools
forget_memoryPermanently delete a specific memory by its key or ID. Destructive operation: the memory cannot be recovered after delet✓ no
index_knowledgeScan and index knowledge base files from the configured Obsidian vault (or other knowledge adapter). Parses markdown fil✓ no
recall_from_knowledgeSearch the indexed knowledge base (Obsidian vault notes, documentation files) using FTS5 full-text search. Separate from✓ no
recall_allUnified search across BOTH personal memories (from recall_memories) AND knowledge base (from recall_from_knowledge) simu✓ no
declare_preferenceExplicitly declare a user preference without going through classification. Bypasses the classifier and directly stores a✓ no
get_memory_profileReturn the complete user identity profile (whoami). Aggregates all stored memories into a structured profile grouped by ✓ no
get_system_promptBuild a complete AI system prompt with relevant memories injected according to context. Selects memories matching the cu✓ no
summarize_and_storeSummarize a long text (conversation transcript, article, document) into concise memory entries and store them. Extracts ✓ no
consolidate_memoriesRun memory consolidation pipeline to clean up and optimize stored memories. Three phases: P0 (deduplication via Jaccard ✓ no
schedule_consolidationSchedule automatic background consolidation to run at a fixed interval using threading.Timer. After scheduling, consolid✓ no
stop_consolidationCancel an active automatic consolidation schedule started by schedule_consolidation. If no schedule is active, returns g✓ no
add_ruleCreate a new behavioral rule that will be automatically matched and injected into future prompts. Rules have: trigger pa✓ no
list_rulesList all stored rules or filter by criteria (status, type, scope, keyword). Returns array of rules with {id, trigger, bo✓ no
match_rulesFind all rules that match a given context string or user message. Evaluates each active rule's trigger pattern against t✓ no
inject_rulesTake matched rules and format them into injectable prompt text. Applies template formatting, scope filtering, and confli✓ no
my_rulesQuick view of the user's currently active rules in human-readable format. Returns a summarized list showing trigger patt✓ no
delete_rulePermanently delete a rule by its ID. Destructive operation: cannot be undone (though rules are typically simple enough t⚠ yes
suggest_rulesAnalyze stored memories to detect repeated patterns and suggest new candidate rules. Looks for: frequently occurring pre✓ no
promote_rulesPromote one or more candidate rules (from suggest_rules output) to active status. Converts suggestions into real rules t✓ no
update_ruleModify an existing rule's properties (trigger, body, scope, type, priority). Partial update: only specified fields are c✓ no
my_profileComprehensive user identity profile combining memories, rules, and preferences into a single structured output. More det✓ no
onboardStart the new-user onboarding flow. Accepts name, role, initial preferences, and optional namespace. Creates initial mem✓ no
⚙ Install config
Source-only — no published npm / pypi package detected.
Clone and follow the build instructions in the repo: github.com/lulin70/carrymem
📈 Score historylast 17 snapshots
5/26/20266/13/2026 · 17 snapshots
⚙ Maintenance health
maintenance data not yet available — check back later.
⛁ Raw data
weekly downloads0
github stars0
forks0
open issues0
license✗ missing
readme length0 chars
last updated19d ago
owner of this server? claim your listing to get a verified badgeclaim →
🔔 Score drop alerts
get notified by email when this server's score drops 5+ points
carrymem — MCP Score: 15/100 | MCPScore | Timeahead