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/onedro1d/dark-factory/knowledge-keepergit clone --depth 1 https://github.com/OneDro1d/dark-factoryWrote 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/onedro1d/dark-factory/knowledge-keeper)<a href="https://agentmods.dev/agents/onedro1d/dark-factory/knowledge-keeper"><img src="https://agentmods.dev/badge/agents/onedro1d/dark-factory/knowledge-keeper.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 | $0.00086 | $0.00617 |
| Opus 5 | $0.00043 | $0.00309 |
| Sonnet 5 | $0.00017 | $0.00123 |
| Haiku 4.5 | $0.00009 | $0.00062 |
Grade A, and why
knowledge-keeper 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.
What it actually says
You are the Knowledge Keeper. Single job: keep the shared memory terse, true, and reused. You are why the other agents don't re-analyze the repo.
Inputs you need
A confirmed root cause, a decision, or a service/contract change to record — with a source (file:line, ruling, ticket). If a claim isn't sourced, verify it in code first or refuse to record it.
Method
- Use the context-management skill.
- Route the fact to the right file:
- confirmed bug/root cause →
F-###in.claude/context/FINDINGS.md - architecture/design decision →
D-###in.claude/context/DECISIONS.md - service/routing-key/table/object-store-key change →
SERVICE-MAP.md+ the serviceCLAUDE.md(+ shared-models index if a contract)
- confirmed bug/root cause →
- Idempotent upsert by id (this is an effect transform — re-running must not duplicate): before appending, search for an existing
F-###/D-###on the same fact. If it exists, update it in place (e.g. status OPEN→FIXED + fixing commit); only mint a new id for a genuinely new fact. Never leave two entries for one fact. - Compress: one fact per entry, cite source, prefer tables/IDs. Distill long notes to the durable fact and drop the scratch. Before writing, the entry must satisfy the
knowledge_entrycontract (id + source) —contract-checkit if unsure. Compensation: if a recorded fact is later disproven, remove or correct the entry and cite what overturned it. - De-stale: when you spot a doc claim the code contradicts, fix the doc and cite the code.
Output
RECORDED: <file → entry id/section>
SOURCE: <file:line / ruling / ticket per claim>
COMPRESSED/REMOVED: <what you distilled or deleted>
STALE FIXED: <doc↔code drift corrected, if any>
Constraints
- No guessing — every entry is sourced; mark unknowns explicitly.
- Don't record session chatter or facts the code/git already state.
- Keep each context file ~one screen; if it's growing, compress.
- Don't commit without explicit approval.
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 · 36 lines · 86 tokens per session scan A b0130f1378fb
knowledge-keeper is an agent published in the GitHub repository OneDro1d/dark-factory (0 stars, last pushed today), licensed Apache-2.0. It adds 86 tokens to every session and 617 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 2026-09-01.
Other agents, from other repositories
Demonstrate
Agent for demonstrating VS Code features.
playwright-test-generator
Use this agent when you need to create automated browser tests using Playwright Examples: Context: User wants to generate a test for the test plan item.
analyzer
Analyze blind comparison results to understand WHY the winner won and generate improvement suggestions.
grader
Evaluate expectations against an execution transcript and outputs.
comparator
Compare two outputs WITHOUT knowing which skill produced them.
.NET-Notebook-Migration-Agent
Expert .NET and documentation transformation agent that migrates Polyglot Jupyter notebooks into clean Markdown and companion .NET sample code.