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 skills/bankszach/document-systems/compose-document-systemnpx skills add bankszach/document-systems --skill compose-document-systemgit clone --depth 1 https://github.com/bankszach/document-systemsWhat 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 | $0.00078 | $0.00635 |
| Opus 5 | $0.00039 | $0.00318 |
| Sonnet 5 | $0.00016 | $0.00127 |
| Haiku 4.5 | $0.00008 | $0.00064 |
Grade A, and why
compose-document-system 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 yesterday.
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 — 65 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Compose Document System
Perform the semantic reasoning, then ask the deterministic MCP to construct a proposal-scoped Expression. Render documents as traceable views of that Expression. Do not begin with a universal prose template.
Workflow
- Identify the document's primary job and intended user.
- Call
list_document_profilesand select one primary profile plus no more than one secondary profile. - Read the supplied sources yourself. Mark source review only as
REVIEWED_BY_CALLER; this remains your assertion, not an MCP finding. - Build explicit SourceRefs, semantic elements, and assertions for consequential values or transitions supported by caller-reviewed evidence.
- Prefer
compile_document_packetto compose, validate, gate, and render once. UseSUMMARYfor an efficient review handoff andFULLonly when the caller needs a portable packet to store or transfer. - Use the separate compose, validate, and render tools only when an existing Expression or staged diagnostic workflow requires them.
- Resolve integrity and semantic-assertion failures. Report readiness warnings rather than filling them speculatively.
- Hand off the packet identity, release gate, artifact, and unresolved human decisions. A ready gate still requires a human decision.
Populate only elements supported by the user's request or cited evidence. Keep unknown fields explicit; do not invent authority, acceptance, or proof.
Boundaries
- The Expression state is always
PROPOSED; this skill cannot promote it. - Treat generated requirements, claims, and procedures as proposals.
- Never infer acceptance from validation, a completed checklist, or polished formatting.
- Do not put acceptance, operative status, views, or validation results inside an Expression.
- Do not create multiple independently maintained versions of the same facts.
- Do not force a multi-profile document when one focused profile is enough.
- If the task is a simple low-consequence note, use ordinary writing instead of this system.
- The MCP writes nothing. A FULL packet is portable but remains
NOT_PERSISTED; save it with ordinary repository tools only when the user explicitly requested project writes.
What ships with it
2 files beside SKILL.md in the same directory: the scripts, references and assets a skill reads on demand. Not counted in the per-session cost; read them before you install if any of them is executable.
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.
- yesterday First seen · 65 lines · 78 tokens per session scan A da3e2cd32e4b
compose-document-system is a skill published in the GitHub repository bankszach/document-systems (0 stars, last pushed 1mo ago), licensed MIT. It adds 78 tokens to every session and 635 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
graph-mutation-plan
Cookbook for composing an applygraphmutations plan — stable entitykey patterns, the canonical label/edge vocabulary, evidence/invalidation/confidence discipline, and a worked example. Load this when building a non-trivial mutation plan.
taiyi-ui-design
TaiyiForge 第 4 阶段 — UI/UX 契约，产出 UI-DESIGN.md。四端通用。.
taiyi-evolve
TaiyiForge 辅助 — 实现后架构与文档同步（architecture-sync）。OpenCode / Claude / Codex / Cursor 通用。.
taiyi-diagram-c4
TaiyiForge 辅助 — 从代码反推 C4 架构文档（Observed/Inferred 分层 · Mermaid 真源）。OpenCode / Claude / Codex / Cursor 通用。.
writing-style
Use for technical communication - GitHub/GitLab tickets, PR/MR descriptions, issue comments, code review comments, commit messages. Direct, brief style with no AI-speak. NOT for README.md, public docs, or blog posts.
steer
Project-profile recalibration. Audits session calibration drift, presents per-cluster evidence for a user verdict, writes the updated project-profile rule, and inscribes a settled-direction clause.