Getting it into your agent
One page per mod, every tool's command on it. A separate URL per tool would split the same page into five that compete with each other.
npx agentmods add agents/novasplace/csm/csm-archivistgit clone --depth 1 https://github.com/NovasPlace/CSMWrote this? Show the measurements
A badge with what this costs and how it scanned, read live from this page, so it follows the numbers instead of freezing them. Markdown for a README, HTML for a documentation site or a project page.
[](https://agentmods.dev/agents/novasplace/csm/csm-archivist)<a href="https://agentmods.dev/agents/novasplace/csm/csm-archivist"><img src="https://agentmods.dev/badge/agents/novasplace/csm/csm-archivist.svg" alt="Measured on agentmods" height="20"></a>What it costs to keep this loaded
Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.
| Model | Per session | Once invoked |
|---|---|---|
| Fable 5.1 | $0.00041 | $0.00354 |
| Opus 5 | $0.00020 | $0.00177 |
| Sonnet 5 | $0.00008 | $0.00071 |
| Haiku 4.5 | $0.00004 | $0.00035 |
Grade A, and why
csm-archivist scanned grade A with 0 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 6d ago.
A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.
Nothing flagged
None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.
What it actually says
You are the CSM Archivist, a careful steward of Cross-Session Memory.
Operating rules:
- Resolve
projectRootto the current workspace root and scope every call to it. Never cross project boundaries. - Always start read-only:
csm_memory_governance_report,csm_memory_dedup_detect,csm_memory_candidate_report,csm_memory_archive_candidate_report, andcsm_recall_quality_report. - Present findings and a concrete plan before any mutation.
csm_memory_mergeis exact-match only: it supersedes duplicates and preserves originals — it never deletes. Perform merges only for the specific candidates the user approves.csm_memory_deleteis destructive. Use it only on an explicit, specific user request, never in bulk from a heuristic.- Never surface or persist secrets, credentials, or sensitive content. The current repository state is the source of truth; memory is evidence and provenance.
Report the outcome plainly: what was reviewed, what changed, and what remains.
What this file has done since we first saw it
Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.
- 6d ago First seen · 18 lines · 41 tokens per session scan A b176fb4fb138
csm-archivist is an agent published in the GitHub repository NovasPlace/CSM (42 stars, last pushed 2d ago), licensed MIT. It adds 41 tokens to every session and 354 once invoked, about $0.0002 per session on Opus 5. A static security scan graded it A with 0 findings. No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-30.
Other agents, from other repositories
kodelyth-memory
Manages the local Kodelyth Memory store — captures patterns and solutions from past sessions, recalls them when relevant, and shapes context for prompt-cache savings. Use PROACTIVELY at session start to recall relevant memories, and at session end to capture what worked. Model-agnostic.
learning
Use this agent when the system needs to learn from past sessions, improve recommendations, or adapt to user patterns. This includes: analyzing session history for patterns, improving agent selection accuracy, learning from user corrections, optimizing workflow recommendations, and building user preference models.…
distiller
Turns a raw work session (sessions/archive/ notes, .jsonl transcripts) into clean notes and lessons in the trunk, or updates existing notes with what is new. Run it after a session worth capturing.
gardener
Tends the trunk (/.c-brain/trunk) — files misplaced notes, deduplicates, guarantees every note is on the MEMORY.md + lessons/INDEX.md map, weaves and repairs [[...]] links, masks secrets. Run it after a work session, or when the tree looks untidy.
architect
Keeps the OVERALL COHESION of the trunk — reads the whole graph topology (links, subsets, similarities) to weave missing links, connect isolated notes, spot disconnected islands and inconsistent placement. The wide view, where the gardener files note by note. Run it periodically, or when the tree starts to fragment.
archivist
Manages the freshness of the trunk — proposes archiving dead weight, never deletes on its own.