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 flonat/flonat-research --skill compile-knowledgegit clone --depth 1 https://github.com/flonat/flonat-researchWrote 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/flonat/flonat-research/compile-knowledge)<a href="https://agentmods.dev/skills/flonat/flonat-research/compile-knowledge"><img src="https://agentmods.dev/badge/skills/flonat/flonat-research/compile-knowledge/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/flonat/flonat-research/compile-knowledge"><img src="https://agentmods.dev/badge/skills/flonat/flonat-research/compile-knowledge.svg" alt="Reviewed on agentmods" width="80" height="20"></a>- NVIDIA SkillSpector warn
SkillSpector: 1 finding, up to medium
These are SkillSpector’s own severities. On a checked sample its high-severity flags on skills were ~96% false positives — a documented command, a public API, a “never do X” rule — so we show them as a caution to read, not a verdict. Why →
- medium Excessive Agency · line 46 Skill enables autonomous high-impact decisions without human-in-the-loop verification. Critical operations (destructive commands, financial transactions, data deletion) should require explicit user confirmation.Fix: Add human-in-the-loop confirmation for destructive, irreversible, or high-impact operations. Never auto-execute commands that modify files, send data, or alter system state.
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 | $0.00056 | $0.02415 |
| Opus 5.5 | $0.00022 | $0.00966 |
| Sonnet 5.5 | $0.00011 | $0.00483 |
| Haiku 4.5 | $0.00006 | $0.00242 |
Grade A, and why
compile-knowledge 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 1mo 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 — 256 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Compile Knowledge
Reads raw inputs from a project and compiles/updates a
knowledge/wiki — a collection of LLM-maintained markdown articles that synthesise everything known about the project's domain.
When to Use
- After a literature review to consolidate findings
- After a series of sessions on a project to build up the knowledge base
- When starting work on a project and wanting to understand the state of knowledge
- Periodically to keep the knowledge wiki fresh
When NOT to Use
- For structural project audits — use an installed research-project audit workflow
- For atlas topic metadata — that stays in the vault
- For session-specific notes — those go in
log/
Architecture
The knowledge wiki lives at <project>/knowledge/:
knowledge/
_index.md ← auto-maintained index with brief summaries
concept-name.md ← one article per concept
concept-name-2.md
The LLM owns this directory. the user rarely edits it directly. The value is in the compilation — connecting dots across sources that no single document contains.
Modes
| Mode | Invocation | Behaviour |
|---|---|---|
| Interactive (default) | compile-knowledge [project-path] |
Phases run normally; mid-run nudges allowed where useful; end-of-run summary requires no confirmation but the user can inspect output and re-run. |
| Autonomous | compile-knowledge [project-path] --autonomous (or -y) |
End-to-end run with zero prompts. All decisions take the documented default. Single end-of-run summary is the only user-facing output. Used by the Saturday wiki-grow cron to refresh project knowledge before the wiki-promotion pass. |
Autonomous-mode defaults (every choice point)
When --autonomous / -y is set, the skill takes these defaults silently:
| Choice point | Default |
|---|---|
| Project path unresolved (CWD not a research project) | Fail with a single one-line error to stderr. Do not prompt for path. |
| No sources found in Phase 1 | Skip remaining phases; emit No sources to compile. and exit clean. |
| Concept granularity uncertain | Prefer 400-600 word articles, specific scope over generic. Err on the side of fewer, sharper concepts. |
| Article exists but scope shifted | Update in place when overlap ≥50% with new finding; create new article only when genuinely orthogonal. |
| Contradictions between sources | Flag explicitly in prose ("Source A says X, Source B says Y") — never silently pick a side. |
| Source can't be parsed (encoding, malformed) | Log to the Gaps section in _index.md; continue with remaining sources. |
| Atlas topic match in Phase 5 is 0 or >1 | Skip silently (already documented in Phase 5.4). |
| Article exceeds 1000 words | Split into two articles by sub-theme; never truncate findings. |
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.
- 1mo ago First seen · 256 lines · 56 tokens per session scan A 83659ad2821c
compile-knowledge is a skill published in the GitHub repository flonat/flonat-research (147 stars, last pushed 20d ago), licensed MIT. It adds 56 tokens to every session and 2,415 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 2026-09-03.
Other skills, from other repositories
ask-agents
Send a task or question to Claude Code and/or Gemini as external subagents and report back what they said. Use when the user asks to ask Claude, ask Gemini, ask both, get a second opinion, split reviewer roles, or compare external agents.
prompt
Turn a rough, dictated, or ambiguous request into a clear, structured task, then carry it out. Use when a request is conversational, messy, underspecified, or high-stakes enough to deserve careful framing before work begins.
review-plan
Review a proposed plan with a single-pass structured critique and a clear verdict. Use when the user asks to review a plan, stress-test a plan, critique a workflow, identify risks, or improve an execution sequence.
fin-paper-convert
Compile LaTeX to PDF and convert to target journal format.
fin-paper-plan
Generate structured paper outline adapted to target journal.
done
Wrap up the current session by writing a short HANDOFF.md in the project folder, so the next session can resume quickly. Use when the user says done, wrap up, close this out, or capture the session.