Skip to content
Skills Library
DevOps & cloud · Review & audit

memory-stats

/memory-stats

Quickly check the health and coverage of the RivetOS memory store. Use when the user asks about memory status, how much history exists, compaction state, or "is memory working?

Repository
rivetos ↗
Licence
Apache-2.0
Last repository change
26.10.2026

Assessment and security

90/ 100Class: good
Format25 / 25
Frontmatter with name and description, valid name, description up to 1,024 characters, name matching the folder.
Discoverability25 / 25
Description explains when to use the skill, has a reasonable length and uses third person.
Content25 / 35
At least 150 words and at most 500 lines, headings, step lists, examples or code.
Hygiene15 / 15
No placeholders such as TODO, no Windows paths, with a licence.
Riskno signals detected in the scan
Files in this package
1
Scripts
0
HTML comments
0
URLs in SKILL.md
0
Deprecation

No signals detected. Repository changed on 26.10.2026, without references to retired models or APIs.

Licence

Apache-2.0, otwarta: można używać i zmieniać, z zachowaniem licencji i informacji o zmianach.

What it does and when to use it

Quickly check the health and coverage of the RivetOS memory store. Use when the user asks about memory status, how much history exists, compaction state, or "is memory working?

Original author description

Quickly check the health and coverage of the RivetOS memory store. Use when the user asks about memory status, how much history exists, compaction state, or "is memory working?

How it works

  1. Step 0 — Whose memory is this? (identity check)
Original SKILL.md headings · 51 lines, 384 words
SKILL.md preview
---
name: memory-stats
description: 'Quickly check the health and coverage of the RivetOS memory store. Use when the user asks about memory status, how much history exists, compaction state, or "is memory working?"'
tags: [rivetos, memory]
version: 0.1.0
---

# Memory Stats

## Step 0 — Whose memory is this? (identity check)

Before recalling anything, check the routed user: `echo "${RIVETOS_USER_ID:-}"`.
One predicate, everywhere:

Then resolve the owner id — one executable step, no reading USER.md:
`cat users/profiles.json 2>/dev/null` and take the reserved `"_owner"` key.

- **Env empty, or equal to that `_owner` value** → you are serving the
  **node owner**. Everything below applies unchanged — and when the env WAS
  set, still name whose memory you searched.
- **Any other id — or the env is set and `users/profiles.json` (or its
  `"_owner"` key) is absent** → treat the session as **routed** to that
  user. This is deliberately fail-safe: a missing map can make the owner's
  own session slightly more formal, but it can never lock anyone out or
  disclose the owner's context to a guest. In routed mode: the memory tools
  already hit _that user's_ database (#561), so findings are their history.
  Address them directly; do not apply the owner's projects, preferences, or
  past to them; and the owner's USER.md / private workspace context is **not
  yours to disclose** to them. Resolve the id to a display name via
  `users/profiles.json` or `users/<id>.md` — if neither exists, use the raw
  id, never guess. Name whose memory you searched in every answer
  ("searched coco's memory for…").

The store your tools hit is shared by every Rivet agent **serving this same
user** — cross-agent recall stays first-class, per user, never across users.

**Delegating to `memory-researcher`? Bind the phrase to the branch** — owner
mode says exactly "node owner"; routed mode says exactly
"routed user: <name> (<id>)". Never phrase the owner as a routed user. The
researcher has no shell and cannot run this check; an unstated identity
forces it into neutral framing.

Call the `memory_stats` (or `rivetos__memory_stats`) tool from the RivetOS MCP server.

This gives a high-level view of:

- Alerts first: stuck graphile jobs, orphan summaries, embedding/compaction backlog
- Number of conversations and messages
- Coverage by agent (`rivet-claude`, `rivet-hermes`, `rivet-grok`, `rivet-kimi`, `rivet-gpt`, `rivet-deepseek`, `rivet-qwen`, etc.)
- Compaction / summarization status
- Any obvious gaps

Read the alert sections before the per-agent census. Those now render first so
truncated MCP payloads still surface silent rot.

Useful as a diagnostic before a big recall session or when debugging why memory searches are returning thin results.

Versions

RepositoryScoreLicence
This page · philbert440/rivetos90Apache-2.0
philbert440/rivetos90Apache-2.0
philbert440/rivetos90Apache-2.0
philbert440/rivetos90Apache-2.0
philbert440/rivetos90Apache-2.0
philbert440/rivetos90Apache-2.0
philbert440/rivetos90Apache-2.0

Database snapshot: 13.11.2026 · package version 0911832261bd (SHA-256 digest of the entire package)

Downloads in the last 30 days: 0 · ranking

Report a problem with this skill: harmful content, incorrect descriptions or an author’s removal request.

Similar skills

  • 100rivetos-status
    Check RivetOS connection health for Claude Code: mode (cloud vs local), Tailscale up?, DataHub/cloud endpoint reachable? Use when the user asks if memory is connected, why recall is empty, or to diagnose onboard. Never dump secrets.
  • 100rivetos-onboard
    First-run and re-runnable RivetOS setup for Claude Code. Use when the plugin is newly installed, memory tools are empty/disabled, RIVETOS_MODE is unset, or the user says set up / connect / onboard RivetOS. One fork: cloud vs local.
  • 100cortex-explore-memory
    Explore the memory system's state, find gaps in knowledge, assess coverage, and get diagnostic information. Use when the user asks 'what does my memory look like', 'show me memory stats', 'what am I missing', 'how good is my knowledge', 'memory health', 'show coverage', 'find…
  • 90review-agent-memory
    Reviews agent memory and state-store configuration for correctness. Use for "review agent memory", "check state store config" or "audit agent persistence".

Feedback and reports

Your message goes to Bartosz Sokoliński. An email address is needed only for a reply. See the privacy policy.