devops-infra MCP Server
Autonomous memory system for AI infrastructure.
Discovered via github-topic:mcp and last synced 3mo ago.
Install instructions not detected yet
Check the source repository for the latest setup steps.
audit log for memory creation, enrichment, merge, demotion, and retirement
master record for archived source material
fact-to-segment support links
opt-in nightly git commit source
Model output from one source only
support and counter-evidence for rules
verify configured vLLM endpoints
no
no
no
provenance for graph edges
yes
active agent or task session state
yes
support and counter-evidence for beliefs
media metadata for screenshots, images, pages, video, and documents
yes
graph-first knowledge extraction
yes
no
large-corpus summarization
Model output confirmed by multiple evidence sources
discrete events inside an episode
typed graph links between facts, memories, episodes, entities, and events
inferred operational belief kept separate from facts
-->
Operator-approved corrections or summaries
time-bounded incidents or sessions
gold-set cases used by the eval path
Purpose
consolidated patterns, heuristics, playbooks, summaries
evidence links for memory revisions
audit trail for extraction attempts
human review and moderation workflow
ordered workflow steps and expected observations
evidence for health decisions
stored evaluation runs and aggregate metrics
no
Best at
synthesized entity state with primary evidence
Purpose
Parsed directly from logs, YAML, or strongly-typed files
migration ledger for applied SQL files and baselined deployments
evidence-grounded reusable operational workflows
scoped retention and staleness snapshot
yes
no
evidence spans, excerpts, embeddings, and scope
memory-to-fact and memory-to-episode support
evidence-linked operational invariant or policy
multi-vector embeddings by target kind and vector role
surface blocked or failed artifacts
service-to-project scoping map
deterministic memory health/risk assessment
derived-memory class registry
Best entry point
structured actor/action/target/cause/effect timeline frames
Input
Required
no
conversational memory and fast hosted adoption
hierarchical memory and reflection loops
What it is
Metric
`benchmarks/results/2026-04-09-retrieval.json`
devices, services, projects, incidents, sessions, config items
9
no
the central truth table with confidence and validity windows
publish-state tracking for promoted memory notes
provenance for entity summaries
expiring session-local memory with explicit promotion status
layout-aware regions linked to grounded objects