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 skills add olegvg/olegvg-skills --skill kb-maintaingit clone --depth 1 https://github.com/olegvg/olegvg-skillsWrote 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/skills/olegvg/olegvg-skills/kb-maintain)<a href="https://agentmods.dev/skills/olegvg/olegvg-skills/kb-maintain"><img src="https://agentmods.dev/badge/skills/olegvg/olegvg-skills/kb-maintain/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/skills/olegvg/olegvg-skills/kb-maintain"><img src="https://agentmods.dev/badge/skills/olegvg/olegvg-skills/kb-maintain.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 | $0.00084 | $0.01334 |
| Opus 5 | $0.00042 | $0.00667 |
| Sonnet 5 | $0.00017 | $0.00267 |
| Haiku 4.5 | $0.00008 | $0.00133 |
Grade A, and why
kb-maintain 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 11d 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 — 107 lines — stays where its author put it; the contents beside it link to each section on GitHub.
kb-maintain — Guided KB Maintenance Flow
When to use
User wants the KB brought current but has not specified whether the work is a drift check, a scoped refresh, decision recording, or a sequence of those operations.
This skill is an orchestrator. It does not implement refresh or decision-record logic directly; it delegates to the atomic skills and keeps each write boundary visible to the user.
Input
- Optional scope path(s), e.g.
src/billing/orsrc/billing/ src/auth/. - Optional plan path, e.g.
docs/plans/2026-03-19-1400/plan-billing-rewrite.md. - Optional mode words like
check only,refresh,record decisions, orfull.
If the user's intent is ambiguous, ask one short clarification before writing anything.
Workflow
1. Branch on user input
If .kb/ does not exist, stop and suggest invoking the kb-build skill (or /kb-build).
Then choose a path:
- Explicit scope provided (
/kb-maintain src/billing/or one or more path arguments) → fast path, see step 2a. Skip drift-check. - Plan path provided (
/kb-maintain docs/plans/.../plan-X.mdor--record <plan>) → record path, see step 2b. Skip drift-check. - No arguments → discovery path, see step 2c. Invoke
kb-check-driftto surface candidates first.
The drift-check is read-only and never modifies .kb/. Skipping it on the fast/record paths is safe: the user already specified what to do.
2a. Fast path — direct refresh
Invoke the kb-refresh skill once with the full list of approved paths. Prefer one call with multiple paths over N sequential calls — see kb-refresh's multi-path input contract.
If the user gave one path, that is one batched refresh of one path. If many, that is one batched refresh of many. Either way: one invocation, one git diff, one state advance.
After the refresh returns, surface any decision candidates the refresh reported and ask whether to also record them (step 2b for each approved plan).
2b. Record path — decision recording
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.
- 11d ago First seen · 107 lines · 84 tokens per session scan A 459fda981699
kb-maintain is a skill published in the GitHub repository olegvg/olegvg-skills (7 stars, last pushed 2mo ago), licensed MIT. It adds 84 tokens to every session and 1,334 once invoked, about $0.0004 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
engram
Read and write the project's durable reasoning memory (decisions, principles, cautions, problems, insights) through the Engram MCP tools. Recall relevant memory before non-trivial work; capture only durable, high-value knowledge silently at natural stopping points; keep the graph honest (judge suspects, close answered…
engram-digest
Seed or top up this project's Engram graph from the existing codebase — an explicit, user-invoked digestion of the current working tree into typed memory nodes (decisions, principles, cautions, problems, insights, intents). Use when the user says "digest this project", "ingest the codebase into memory", "seed the…
bootstrap-llm-synthesis
Construct the LLM synthesis prompt from project surface scan + optional tree-sitter context + optional Q&A answers. Call the LLM. Parse and validate the response into 6-8 structured memory entries with clarity tags and source traceability. Used as Stage 3 of the /gaai:bootstrap pipeline.
memory-delta-triage
Apply three deterministic heuristics to a single memory-delta file to produce a structured verdict block; invoke memory-ingest on ACCEPTED candidates only in validate mode. Activate when Discovery processes a raw memory-delta from contexts/artefacts/memory-deltas/.
memory-reconcile
Scan all memory files, documentation (/docs//.md), and README files (/README.md) for drift, contradictions, and stale references. Produce a reconciliation report for Discovery to action. Activate on demand or via cron.
memory-archive-superseded
Migrate a superseded DEC's index rows from active index.md to archive/superseded-decisions.archive.md. Idempotent. Discovery-only — never invoked by daemon delivery. Updates DEC frontmatter as canonical source of truth.