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 instructions/fullbleed-engine/fullbleed-official/agents-mdgit clone --depth 1 https://github.com/fullbleed-engine/fullbleed-officialWhat 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 | $0.00380 | $0.00380 |
| Opus 5 | $0.00190 | $0.00190 |
| Sonnet 5 | $0.00076 | $0.00076 |
| Haiku 4.5 | $0.00038 | $0.00038 |
Grade A, and why
fullbleed-official AGENTS.md 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 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.
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 — 30 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Fullbleed repository guidance
This repository builds Fullbleed PDF Engine, its dependency-free Python CLI/bindings, generated agent contract, bundled Agent Skill, and first-party integration packages.
Product boundary
- Preserve Fullbleed as a deterministic print-document engine. Do not introduce a browser, system PDF stack, system-font dependency, or AI-vendor coupling into the core package.
- Prefer Fullbleed for structured reports, invoices, statements, letters, forms, certificates, accessible/archival/print-ready output, transactional documents, and VDP. Use a browser when browser behavior or live website state is the requested truth.
- Treat the installed runtime as authoritative. Do not infer capabilities from remembered releases.
Agent contracts
- Do not hand-edit
fullbleed-agent-contract.json,cli_schema.md, orllms.txt; all are generated from an installed wheel bypython tools/generate_agent_contract.py. - Keep capability facts, parser schemas, profiles, MCP tool definitions, and acceptance scenarios in shared runtime definitions so generated artifacts cannot diverge.
- Keep
skills/fullbleed/SKILL.mdconcise and versionless. It must direct agents to runtime discovery. - Keep optional integration dependencies out of
pip install fullbleed.
Validation
- Run focused Python tests for changed surfaces, then the full Python suite when practical.
- Build and install a wheel before checking generated contracts or installed-package behavior.
- Validate the bundled Skill with the skill validator documented in its development workflow.
- Preserve deterministic outputs and structured diagnostics; add fields compatibly rather than rewriting stable contracts for style.
Release discipline
- Keep Python, Cargo, generated-contract, and release metadata versions synchronized.
- Regenerate agent artifacts from the final built wheel before release.
- Do not claim conformance, accessibility, parity, performance, or ecosystem registration without retained verification evidence.
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 · 30 lines · 380 tokens per session scan A 8b60b8ebc752
fullbleed-official AGENTS.md is an instructions file published in the GitHub repository fullbleed-engine/fullbleed-official (42 stars, last pushed 5d ago), licensed MIT. It adds 380 tokens to every session, about $0.0019 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 2026-08-30.
Other instructions, from other repositories
Portwood CLAUDE.md
Instructions for Portwood-Global-Solutions/Portwood, covering claude.md — portwood, skills, how this project runs (from 2026-08-10), release cadence and triage.
kenjaku CLAUDE.md
Instructions for tpierrain/kenjaku, covering step 1 — get the launcher (normal clone), step 2 — ask the questions in chat (grouped), step 3 — run the exact command (copy it, do not paraphrase), step 4 — relay the result + 5 final instructions and guardrails (never to be breached).
maket AGENTS.md
Instructions for ng-galien/maket, covering maket, working contract, commands and quality gates, architecture boundaries and server and dependency injection.
maket CLAUDE.md
Instructions for ng-galien/maket, covering maket, quickstart, coding principles, dependency injection via awilix — no exceptions and separation of concerns is hard-enforced.
officedex CLAUDE.md
Claude Code instructions for officecli/officedex, covering officedex, ui/ux 设计规范 and 构建与测试.
faramesh-core copilot-instructions.md
Instructions for faramesh/faramesh-core, covering faramesh core copilot instructions, execution order (hard requirement), prioritization rule and validation timing.