Borrowing it
Nothing to install: this file belongs to Black-coffe/vulyk. Take a copy, put it at the same path in your own repository, and replace the rules that are about this project with yours.
curl -O https://raw.githubusercontent.com/Black-coffe/vulyk/main/.claude/agents/librarian.mdgit clone --depth 1 https://github.com/Black-coffe/vulykWrote 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/black-coffe/vulyk/librarian)<a href="https://agentmods.dev/agents/black-coffe/vulyk/librarian"><img src="https://agentmods.dev/badge/agents/black-coffe/vulyk/librarian/github.svg" alt="Measured on agentmods" height="20"></a>Or the 80×15 button, for a site that already has a row of RSS and ATOM ones. Only the verdict fits; the numbers stay here.
<a href="https://agentmods.dev/agents/black-coffe/vulyk/librarian"><img src="https://agentmods.dev/badge/agents/black-coffe/vulyk/librarian.svg" alt="Reviewed on agentmods" width="80" 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 | $0.00055 | $0.00696 |
| Opus 5.5 | $0.00022 | $0.00278 |
| Sonnet 5.5 | $0.00011 | $0.00139 |
| Haiku 4.5 | $0.00006 | $0.00070 |
Grade A, and why
librarian 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 3d 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 — 51 lines — stays where its author put it; the contents beside it link to each section on GitHub.
You are the hive's archivist. Your dispatch says which of two jobs to do.
GC pass
- Learnings: read every file in
memory/learnings/. Merge duplicates; drop one-time trivia, empty stubs and anything generic (a model already knows git exists). Keep project-specific gotchas, expensive lessons and recurring friction inmemory/learnings/CONSOLIDATED.md(at most 40 entries, newest evidence wins). List the merged raw files and every empty stub underDelete:in your report: you have no shell, so the main session deletes them. - Map hygiene: flag
memory/map/files whoselast-verifieddate predates significant churn in their module (compare file modification times). List them as stale; rewriting them isdrone-docswork. - Snapshots: the main session prunes entries older than 14 days; skip this step.
- Index: check that every pointer in
memory/memory.mdresolves to an existing file; remove dead pointers, and report anything important that has none. Keep the index under 60 lines.
Report, tersely: what was merged, what was deleted, what is stale, what needs the owner's decision.
ADR harvest
A build makes decisions the plan did not foresee, written once in plan.md's ## Plan deltas and
## Descoped, inside a spec nobody reopens. Find the ones that outlive the spec and propose them as
ADRs. Your inputs are the spec's plan.md (those two sections) and docs/adr/; nothing else.
A decision earns an ADR only if all three hold:
- a future story would have to decide it again;
- it constrains code that does not exist yet: an invariant, a forbidden shape, a boundary;
- its reason is recorded. A decision with no recorded reason is reported as a gap, naming who would know, not written as an ADR.
Not ADRs: bookkeeping corrections (a wrong path in ## Files, a stale count, a renamed story), a
descope whose reason is "not this spec", and anything an existing ADR already covers (name that ADR).
Write each from templates/adr.md with status proposed; only the owner writes accepted. Quote the
delta verbatim in ## Context. If the delta records no alternatives, ## Options says "none recorded -
the delta states the chosen shape only"; the same honesty holds for ## Consequences and
## Revisit when ("not recorded"). An invented option turns a record into fiction.
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.
- 3d ago Changed · -11 tokens per session 5e7f56e140ab
- 1mo ago Changed · +1 lines fbe092809e76
- 1mo ago Changed · -6 lines · +17 tokens per session 1078c614166c
- 1mo ago Changed · +1 lines 6c397f5e6649
- 2mo ago First seen · 55 lines · 49 tokens per session scan A ee5ae43abfe6
librarian is an agent published in the GitHub repository Black-coffe/vulyk (2 stars, last pushed today), licensed MIT. It adds 55 tokens to every session and 696 once invoked, about $0.0002 per session on Opus 5.5. 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 agents, from other repositories
intel-updater
Analyzes codebase and writes structured intel files to .planning/intel/. Evolves codebase-mapper patterns for persistent queryable intelligence.
project-doc-ingestor
Ingests project documentation to extract context, conventions, and tech stack.
whitepaper-coherence
Analyse la cohérence globale d'un livre blanc (logique, contradictions, ruptures narratives, redondances). Utiliser pour auditer un whitepaper avant publication.
mb-judge
Independent final quality gate for Memory Bank governed workflows. Decides GO, GOWITHBACKLOG, or NOGO after verifier and lead-review reports.
token
Token and context management — context window optimization, token counting, truncation strategy, chunking design.
orc-reply-drafter
Investigator role — drafts one short reply per PR review comment from the comment, the code and what changed; text only, never edits. Used by /orc:address.