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 skills/plumbkit/plumb/plumb-memorynpx skills add plumbkit/plumb --skill plumb-memorygit clone --depth 1 https://github.com/plumbkit/plumbWhat 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 | $0.00046 | $0.00917 |
| Opus 5 | $0.00023 | $0.00458 |
| Sonnet 5 | $0.00009 | $0.00183 |
| Haiku 4.5 | $0.00005 | $0.00092 |
Grade A, and why
plumb-memory 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 2d 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.
How it starts
The opening of the file, as written. The whole thing — 55 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Plumb keeps per-workspace memories as markdown under <workspace>/.plumb/memories/, indexed for ranked search. They are for knowledge that outlives one session and has no better home in the code — not a scratchpad, and not a substitute for a comment or a doc.
Under a lean tool profile only search_memories is advertised; set [tools] profile = "full" in .plumb/config.toml to get the rest of this lane.
1. Look before you derive
Re-deriving a constraint someone already recorded is the failure this lane exists to stop.
search_memories(pattern="rate limiter")
relevant_memories(path="internal/tools/edit_file.go")
search_memories is ranked FTS5 while the index is fresh and falls back to substring grep otherwise. relevant_memories takes the other route in — the memories whose stored paths globs match a file you are about to touch, which is the one you want immediately before an edit.
session_start already surfaces the top memories and any last-session summary, so re-listing them is usually wasted; reach for these when you need something it did not show.
2. Pick the search mode deliberately
search_memories(pattern="essio", mode="grep")
mode is auto (the default), fts, or grep. FTS5 tokenises whole words, so a substring inside an identifier — essio within UserSession — never matches it. Auto notices an empty FTS result and greps; a literal fts keeps the empty answer. Setting case_sensitive to true also forces grep, because FTS5 is case-insensitive; leaving it unset keeps the default smart-case behaviour. use_regex forces grep too.
3. Read the one you want, whole
list_memories()
read_memory(name="topology-resync-pacing")
read_memory prints a provenance footer for a memory that is not user-authored. Weigh it: confidence is one of user, generated, imported, or inferred, and only user means a person wrote it deliberately. generated covers both the daemon's rule-based idle summaries (named episodic-*) and anything an agent handed over with share_findings (named finding-*) — a name prefix, not a tier of its own. Ranking and hint injection already prefer user; your reading should too.
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.
- 2d ago First seen · 55 lines · 46 tokens per session scan A fadf3f9266c8
plumb-memory is a skill published in the GitHub repository plumbkit/plumb (4 stars, last pushed 2d ago), licensed MIT. It adds 46 tokens to every session and 917 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 2026-08-31.
Other skills, from other repositories
nim-mcp-tools
Use for Nim symbol navigation, diagnostics, and type resolution. MANDATORY: Use specialized MCP tools (nimFindSymbols, nimFindReferences, nimListSymbols, nimCheckFile, nimCheckProject, nimFindTypeDefinition) first; fall back to grep only on error or user confirmation.
yaml
Validate and understand schema-backed YAML (Kubernetes, Flux, and JSON-Schema configs) with the yayamlls language server. Use when editing or reviewing .yaml/.yml files, diagnosing schema validation errors, configuring .yayamlls.yaml, suppressing a diagnostic, or working with Flux HelmRelease / Kustomization rendering.
lsp-config
This skill should be used when the user asks about "LSP configuration", "language server setup", "adding LSP to Claude Code", "mason config import", "lspconfig migration", "lspctl plugin", or discusses Claude Code LSP integration and troubleshooting.
interview
Ask one useful structured question at a time only when material product/implementation choices are genuinely missing; remember answers and produce a brief/spec. Discoverable facts should be investigated instead of asked.
writing
将共享历史中的已验证事实和计算结果整理成符合受众、格式与长度约束的成稿。.
test
Detect the project’s test stack, run the narrowest useful tests, create tests when authorized, and report coverage/gaps honestly.