PMB AI MCP server

Read from the package source without running it. The installed server may list more.

• apply_daily_decay

  Apply forgetting curve to all events.

• arc_detail

  Get full detail of one arc: title, summary, member events.

• auto_resolve_conflicts

  Auto-archive obvious 'supersede' conflicts.

• chain_current

  Latest milestone of a chain - the 'current state'. Use to answer 'how many X do we have now?' / 'what's the latest?'.

• chain_history

  Full chronological history of a state-chain. Reconstructs the evolution: state_1 → state_2 → ... → state_N with the reason at each step.

• clear_predictive_cache

  Clear all entries from the predictive cache.

• cluster_into_arcs

  PMB v2 Phase 3: LLM groups events into narrative arcs (story threads). Arcs help recall on 'tell me about X' style queries.

• cold_stats

  Get info about cold storage (archived events database).

• compact_storage

  Compact storage: move old archived events to cold storage and VACUUM.

• consolidate_recent
• dedupe_list_pending

  List borderline duplicate pairs awaiting verdict (L2.5). These are pairs that scored cosine 0.80-0.92 - too risky to auto-merge, but worth review.

• dedupe_run_pendingwrite action

  Drain the borderline pair queue via LLM verify. backend: 'auto' (try Ollama then Anthropic) | 'ollama' | 'anthropic'

• dedupe_sweep

  Improvement U: one-shot dedup over all active events.

• dedupe_undo

  Restore events archived by previous dedup runs. Reversible.

• detect_conflicts

  Find conflicting facts between different timestamps.

• extract_facts

  Improvement D (mem0-style): LLM extracts atomic facts from recent events. Each fact becomes a new searchable event.

• feedback_summary

  Aggregate user recall feedback. Returns useful_rate and top flagged-wrong events. This is the real-user health metric.

• file_correlations

  Find files that are often modified together with the given file.

• file_history

  Get commit history for a specific file.

• find_lessons

  Pull procedural lessons (project rules / gotchas) relevant to a topic. Use this BEFORE making a project-shaping choice - picking a library, setting up tooling, choosing an approach - to see what worked or failed before.

• forget

  Archive a memory. Not deleted, can be restored.

• get_subfacts

  Return all subfacts linked to a parent event (Improvement P).

• graph_neighbors

  Co-occurrence neighbors of an entity. Useful for 'what relates to X'.

• graph_stats

  Counts of entities and edges in this workspace's association graph.

• graph_top_entities

  Most-mentioned entities. kind = 'file' | 'tech' | 'concept' or None.

• health_trend

  Get trend of self-test accuracy over time.

• index_pdf

  📄 Extract text from a PDF and persist it as searchable memory.

• index_project

  📂 Index a code project's structure - per-file symbols, imports, languages - so the agent can recall things like "where is the auth flow", "which files import LanceDB", "show me the recall pipeline".

• lesson_impact

  📊 Earned Memory: which lessons actually HELP outcomes (not just which were read or followed). Joins each surfaced lesson to the OUTCOME of the turn it was active in - tests pass/fail, red->green, build, deploy (no LLM) - and returns per-lesson success_rate, lift vs the no-lesson baseline, and churn. Use to spot dead-weight or HARMFUL rules (negative lift). Worst lift first.

• list_arcs

  List narrative arcs in the current workspace.

• list_goals

  List goals (optionally filter by status).

• list_recent

  List the most recent active events in the workspace.

• mark_lesson_followed

  Confirm whether a previously surfaced lesson actually changed your behaviour on the current task. Call this AFTER acting on a lesson - the self-improvement loop uses this signal to identify useful vs dead lessons.

• overview

  Get a structured OVERVIEW of everything memory knows about a topic.

• pin

  Pin a memory: max importance, never auto-archived.

• precompute_predictive_cache

  Improvement F (PMB-unique): during sleep, LLM predicts likely questions and pre-runs recall. Future near-identical queries get instant (~3ms) responses via cosine match.

• prepare

  ⭐⭐⭐ Call this ONCE at the START of any substantive user task.

• project_overview

  ⭐ Use at the START of any project work - returns the FULL project context in ONE call, sourced from the entity graph (not recall):

• project_structure

  🗺️ Pull a project's STRUCTURE straight from memory - no filesystem scan.

• recall

  Search PMB long-term memory. CALL THIS FIRST when user asks about: - past decisions (why did we choose X, what port did we set) - project history (when did we, what happened with) - people (who is X, what did Alice say) - any fact the user might have recorded earlier

• recall_deep

  DEEP recall - explicitly opt into the slow, thorough path.

• recall_exploration

  💡 BEFORE re-exploring the codebase, check if a PAST session already figured this out. Returns memoized conclusions matching `intent`, each with a freshness check: • `fresh: true` -> every source file is unchanged, trust the conclusion and skip the re-reading entirely. • `fresh: false` -> `stale_files` lists what changed since; re-check ONLY those, reuse the rest.

• recall_smart

  Auto-escalating recall, but FAST and deadline-bounded. Starts with a normal local recall; if confidence is low it escalates only with cheap local steps (wider pool + a local rerank when that model is already warm). It NEVER blocks on an LLM / Claude CLI on this path and always returns within `recall.smart_deadline_ms` (default 15s), returning the best result found so far if the budget is hit.

• recent_activity

  Improvement Q: working memory dump - instant, no search.

• record_activity

  Improvement Q: log a working-memory activity.

• record_batch
• record_code

  Record code content. For Python, AST entities (functions, classes, imports) are auto-extracted into the graph layer (Improvement J).

• record_exploration

  💡 AFTER researching something in the codebase (reading/grepping several files to reach a conclusion), memoize it so a FUTURE session can reuse the conclusion instead of re-deriving it.

• record_fact

  STORE a fact for long-term recall. CALL THIS PROACTIVELY whenever the user mentions:

• record_fact_tree

  Improvement P: store a main fact + multiple linked sub-facts in ONE call.

• record_goal

  Improvement R: create a goal/intent (12-th semantic layer).

• record_image

  Record an image (screenshot/diagram) in memory (Improvement J).

• record_keyed_fact

  ⭐ Use for SINGULAR personal attributes that CHANGE over time.

• record_milestone

  Improvement R: record a milestone in a named state-chain.

• record_recall_feedback

  Record real recall feedback for an event.

• reflect_batch

  PMB v2 sleep operation: LLM reflects on recent un-reflected events.

• remember

  Store a Q/A interaction in memory.

• run_self_testwrite action

  Quantify memory health: system asks itself questions from old memories.

• search_images_by_text

  Cross-modal image search. Encodes the query via CLIP text encoder and finds nearest stored image embeddings.

• session_brief

  Re-orient in a long session: a digest of what was decided / done / learned so far THIS session.

• session_current

  Get current active session info (or None).

• session_end

  End the current session. Returns the closed session's info.

• session_start

  Start a new memory session. Used to group related events.

• session_timeline

  Chronological dump of CURRENT session events. Use for post-mortem 'summarize this session' style questions.

• stats

  Get workspace and memory statistics.

• sync_git

  Capture recent git commits into memory.

• update_goalwrite action

  Update goal status / progress. Creates a goal_update event so the full history of changes is preserved.

• what_just_happened

  Quick: last N events of any type from current session. Use to answer 'what were we doing?' / 'recap of this session'. Instant, no search.

• workspace_info

  Identify the current PMB workspace.

• Code scan172 source files scanned8/25
• –Live reliabilityno gateway calls yet and no remote to proben/a
• –Tool poisoningtools not inspected (local package is not executed); not countedn/a
• Auth qualitylocal package, no credentials required12/15
• Maintenancelast push 3 days ago15/15
• Maintainer identityregistry namespace matches repository owner; GitHub account older than a year8/10

Safety score

66/100

higher than 23% of 48,255 servers

Tools

69

the typical server has 7

Tools that change things

4%

typical: 0%. Fewer write tools means less an agent can break

GitHub stars

280

more than 95% of 28,688 servers

Downloads a week

131

more than 68% of 12,191 servers with a package

PMB AI: common questions

Is PMB AI MCP server safe?

With care: it is graded C, so read the findings first (66/100). Read the PMB AI safety report

How do I install PMB AI?

It runs on your machine. Copy the Claude Code, Cursor, VS Code or Claude Desktop config from the install section.

Does PMB AI need an API key?

Not as far as the registry entry and our scan can tell: no credentials are declared or required.

Is PMB AI maintained?

The last commit was 42 days ago (2026-09-23). The latest release is v1.1.0.

What can I use instead of PMB AI?

Servers from other publishers that do the same job: Deja Vu MCP server, Memorix MCP server and Nautilus Compass Memory MCP server.

How does PMB AI compare with other MCP servers?

safety score 66/100 (higher than 23% of 48,255 servers); tools 69 (the typical server has 7); tools that change things 4% (typical: 0%. Fewer write tools means less an agent can break). Full comparison.