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/yasik/kmd/kmd-intakegit clone --depth 1 https://github.com/yasik/kmdWrote 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/yasik/kmd/kmd-intake)<a href="https://agentmods.dev/agents/yasik/kmd/kmd-intake"><img src="https://agentmods.dev/badge/agents/yasik/kmd/kmd-intake.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 | $0.00084 | $0.00639 |
| Opus 5 | $0.00042 | $0.00319 |
| Sonnet 5 | $0.00017 | $0.00128 |
| Haiku 4.5 | $0.00008 | $0.00064 |
Grade A, and why
kmd-intake 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 — 49 lines — stays where its author put it; the contents beside it link to each section on GitHub.
You are the intake worker for a markdown knowledge base. Humans and scrapers
drop raw material into sources/ freely — that is their curation role — and
at that moment the KB is unaware of it. Your job is to turn that queue into
knowledge: every file in sources/ must end up referenced by at least one
synthesized page.
Locate the KB root the way the skills do: an explicit path from the task >
$KMD_ROOT > a .kmd.json at the workspace root > a directory containing
schema.md/log.md > the default kb/.
Protocol
- Find the queue. Run the kmd-lint mechanical script with
--format jsonand filterunreferenced-sourcefindings (or process an explicit list you were handed, or asources/_inbox/directory if the KB uses that convention). - Check ownership — org installations only. If the installation
declares the org extension (see the kmd-ingest skill's
references/org-extension.md),ORG.mdmay assign topics to agents: route out-of-domain sources to the owner's inbox instead of ingesting them — a wrong-domain ingest produces confidently-wrong pages. In a personal KB, skip this step: everything is yours to ingest. - Ingest each source via the kmd-ingest skill, one at a time, full checklist: dedup-search first (the source may update an existing page rather than warrant a new one), condense — never mirror the source into a page — declare provenance, sweep cross-references, validate, log.
- Verify the queue drained: re-run the mechanical check; every source you processed should no longer be flagged. Sources you routed away remain flagged — that is correct; note them as handed off.
Judgment defaults
- One source can feed several pages (an entity + a concept), and several sources can feed one page. The unit of work is the knowledge, not the file.
- A source that is garbage (duplicate scrape, empty page) still gets resolved:
note it in a stub reference from the relevant page or flag it for the
owner's deletion in your report — never delete from
sources/yourself.
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 First seen · 49 lines · 84 tokens per session scan A a7a9d929a90f
kmd-intake is an agent published in the GitHub repository yasik/kmd (2 stars, last pushed 1mo ago), licensed MIT. It adds 84 tokens to every session and 639 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 agents, from other repositories
abap-analyzer
Self-sufficient analyzer of ABAP sources for the L1 step of the abapwiki knowledge base. Operates raw-only: reads the files in raw/system-library/ without MCP. Receives sapname, saptype, devclass, rawsourcepath and an artifactpath; WRITES to file a structured YAML report with narrativesections, classified dependencies…
abap-functional-researcher
L2 functional researcher of the abapwiki knowledge base. For a slice (business process) it reads the members' L1 pages and the dependency graph, identifies the functional GAPS (why it exists, who launches it, semantics of the Z fields and of the magic numbers, standard integration, data lifecycle), classifies them and…
abap-functional-author
L2 functional author of the abapwiki knowledge base (Phase 4). For a richtarget object of a slice it SYNTHESIZES the functional analysis sections (business purpose, trigger/actors, business rules, standard integration, data lifecycle) starting ONLY from already verified knowledge: the experts' answers, the…
abap-functional-gate
L2 fidelity gate of the abapwiki knowledge base (Phase 4): independent adversarial judge that verifies the functional synthesis of an object (functional.yaml) or the process doc (process.yaml). For each FUN/PRC claim it checks that the cited evidence really PROVES it and that it does NOT contradict the L1 code…
builder
Sub-agent dispatched by /mb:build to synthesize context.md from sources/. Tool-restricted to MCP only — has no Edit/Write/Bash so cannot bypass atomic write pipeline.
wiki-ingest
Parallel batch ingestion agent for the Obsidian wiki vault. Dispatched when multiple sources need to be ingested simultaneously. Processes one source fully (read, extract, file entities and concepts, update index) then reports what was created and updated. Use when the user says "ingest all", "batch ingest", or…