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 commands/optimeta/paideia/analyzegit clone --depth 1 https://github.com/OPTIMETA/PAIDEIAWhat 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 | $0.00020 | $0.07434 |
| Opus 5 | $0.00010 | $0.03717 |
| Sonnet 5 | $0.00004 | $0.01487 |
| Haiku 4.5 | $0.00002 | $0.00743 |
Grade C, and why
analyze scanned grade C with 1 finding 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 2d 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.
Tells the agent never to refusehighAnti-refusal
Suppressing the ability to decline removes a core safety control; a later harmful request then succeeds.
- Do not refuse or stop partway — always return what you found, even if partial. How it starts
The opening of the file, as written. The whole thing — 387 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Output language
Read INTERFACE_LANG from .course-meta (default en). All user-facing prose — chat output and narrative parts of the generated index MDs — must be in that language. Keep in English regardless: file paths, slash command names, pattern IDs (P1..Pk), tier markers (🔥🔥/🔥/🟡/⚪) and the ⚠weak flag, § / Ch section anchors, and table column headers (Problem, Primary §, Secondary §, Patterns, HW coverage, Exam tier, etc.) — weakmap, hwmap, and quiz regex on them.
Load skills/course-builder/SKILL.md.
Arguments: $ARGUMENTS
Non-flag tokens (comma-separated, excluding any -- prefixed tokens) are the user's declared weak zones. Tokens beginning with -- are control flags parsed in Step 0.5 below — exclude them from the weak-zone list.
Prerequisite check: verify that converted/ contains files. If empty, tell the user to run /ingest first.
Follow the course-builder Phase 2 analyze pipeline:
Step 0 — Discovery & fan-out plan
List all files matching converted/lectures/*.md, converted/textbook/*.md, converted/homework/*.md, and converted/solutions/*.md. Count the total as N.
Output in $INTERFACE_LANG (keep the token identifier verbatim):
Analyzing N files (0/N)...
Fan-out (mandatory — single-pass over the full converted directory is forbidden):
Spawn one general-purpose Task sub-agent per file, in parallel, up to the workflow concurrency ceiling (currently ~10 parallel agent slots) at once; batches sized to that ceiling. If N exceeds the ceiling, process in sequential batches, waiting for each batch to complete before launching the next — maximize parallelism within each batch. Each agent reads only its own single file and returns a partial index (structured summary) — it must not re-read or transcribe the full original text back to the parent.
First-batch cap (mandatory — batch-1 must provably commit inside the standard window): The first batch is intentionally smaller than the full ceiling. Set batch-1 size = min(FIRST_BATCH_CAP, N), where FIRST_BATCH_CAP is 3–4 files. This ensures that fan-out + Reduce + all three .partial writes + all three renames provably complete and commit a valid course-index inside the standard window — batch-1 must provably commit a valid course-index inside the standard window; subsequent batches widen to the ceiling (~10). Choosing a ceiling-sized first batch risks a SIGTERM (exit 143) with zero committed files if the Reduce + write phase is reached near the window boundary (as observed in FND-002: "assembling section headers" → SIGTERM, 0 files committed). The small first batch eliminates this risk while the subsequent batches widen to recover throughput.
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.
- 2d ago First seen · 387 lines · 20 tokens per session scan C 78d8faeaced9
analyze is a command published in the GitHub repository OPTIMETA/PAIDEIA (91 stars, last pushed 1mo ago), licensed MIT. It adds 20 tokens to every session and 7,434 once invoked, about $0.0001 per session on Opus 5. A static security scan graded it C with 1 finding (tells the agent never to refuse). No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-30.
Other commands, from other repositories
fieldtheory
Explain and drive the Field Theory CLI — bookmark-sourced seeds, repo-aware possibility runs that score ideas onto a 2x2 grid, and the interconnected .md files they leave behind. Trigger when the user asks about ft, bookmarks, seeds, ft possible (or the older name ft ideas), grids, dots/nodes, or how to turn saved…
book-characters
你是一位资深的人物塑造分析专家，精通角色设计、人物弧光、群像塑造等技巧。你的任务是深度剖析小说中的人物创作手法，帮助作者学习如何塑造鲜活立体的角色。.
learn
Append a lesson to an agent's learnings journal (.claude/learnings/ .md).
recall
Search past session observations and lessons for relevant context. Wrap the memorysmartsearch and memorylessonrecall MCP tools.
tree
Show file tree with sync status indicators showing which files are indexed, modified, new, or deleted.
repo-overview
Generate a one-paragraph overview of the current repository.