kb-graph
Provides a self-tending knowledge base for AI agents, automatically harvesting session transcripts into facts and state notes, and injecting relevant context into new sessions via MCP hooks.
Emerging. No concerning findings. Grades remain capped until the project builds reputation through adoption. Grades reflect the full trust pyramid: code, verification depth, and reputation. New projects cap at C until adoption is earned.
How we verified
Verified is a snapshot. Live keeps it current, and builds your track record.
⚡ Connect GitHub → continuous verification on every pushwhy connect →Who stands behind it
uttambharadwaj
Source: Glama
Claim it to get a verified publisher badge, a free copy of our full audit findings, and direct contact for any high-priority issues we find. Or connect your repo for our deepest verification, Live Monitored: read-only, revoke anytime. What we access →
Install from
The grade above is for the source repository. Registries can serve a different version, so we mark the ones we were not able to read.
BETTER_AUTH_SECRETOAuth for remote accessBETTER_AUTH_URLOAuth for remote accessCLASSIFY_MODELModel for write-time AI workCLAUDE_BUS_ROLECLAUDE_PATHClaude CLI binary, used by harvest/classificationKB_API_KEY_CLAUDEAPI keys for remote REST accessKB_API_KEY_GEMINIKB_API_KEY_OPENAIKB_BATCHKB_BUS_DB_PATHKB_BUS_HOMEpointing it somewhere disposable takes KB_DIR and . KB_DIRKB_BUS_SENDERKB_CONTROL_SOCKET_PATHKB_CORS_ORIGINSKB_DIR/trigger-hook-enabled exists. A one-way command can instead be madeKB_EMBEDDING_CACHE_DIRKB_HARVEST_FACTS1/true/yes makes the nightly harvest extract facts as well as lessons. Off because it is the expensive half and writes an open vocabulary unattended. Scheduled jobs inherit no environment, so kb setup copies this into the job definition — set it before setup, or re-run setup after changing itKB_HARVEST_SDK_SESSIONS1/true/yes harvests print-mode (SDK) transcripts too. Off because the harvest's own claude -p calls look like sessions — 98% of candidates on a busy install. Turn on if you drive Claude Code headlessly and want that work capturedKB_HOOK_DAEMON_TIMEOUT_MSKB_PASSWORDDashboard login passwordKB_PORTHTTP server portKB_REPO_ROOTSColon-separated absolute paths searched to verify a verified-tier commit sha or file-path reference. The server's cwd is often a workspace directory sitting one level above every git repo, where nothing ever resolves — set this to that workspace and each immediate subdirectory that is a git repo is searched tooKB_REVIEW_TIMEOUT_MSKB_SHIM_PROBE_TIMEOUT_MSKB_SHIM_RECONNECT_DELAY_MSKB_SHIM_RECONNECT_MAX_DELAY_MSKB_SLOW_HOOK_MSKB_TICKET_REGEXWorkstream autobind: regex that recognizes ticket ids in directory/branch names. Full match (lowercased) becomes the bus channel name. The default deliberately accepts any short prefix so autobind works unconfigured; it will also match same-shaped directory names like node-22, so set this to something exact if that bothers youMAX_THINKING_TOKENSOBSIDIAN_VAULT_PATHVault path (any markdown directory)STATE_MODELSYNTHESIS_MODELDependency freshness
1/14 production deps stale: gray-matter@2023-07-12 (3.2y)
Claim the listing to review these findings one by one and send us a correction where you disagree, straight to the team. Claiming also means we tell you when the grade moves, and reach you first if we find anything urgent.
[](https://m8ven.ai/mcp/uttambharadwaj/kb-graph)?variant=verified to the badge URL.Vetting this one by hand? Tool Check is an MCP that scores other MCPs. Add it once and ask Claude, ChatGPT, or any MCP client to grade a server, surface CVEs, check the publisher, and suggest safer alternatives — before you install.
https://m8ven.ai/api/mcp/tool-check