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/gixsy95/abap_wiki/agents-mdgit clone --depth 1 https://github.com/Gixsy95/abap_wikiWrote 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/instructions/gixsy95/abap_wiki/agents-md)<a href="https://agentmods.dev/instructions/gixsy95/abap_wiki/agents-md"><img src="https://agentmods.dev/badge/instructions/gixsy95/abap_wiki/agents-md.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 5.1 | $0.03283 | $0.03283 |
| Opus 5 | $0.01641 | $0.01641 |
| Sonnet 5 | $0.00657 | $0.00657 |
| Haiku 4.5 | $0.00328 | $0.00328 |
Grade A, and why
abap_wiki 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 6d 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.
Copies of this mod
1 near-identical copy found in the catalogue:
- abap_wiki CLAUDE.md — 97% identical, 4 lines differ
How it starts
The opening of the file, as written. The whole thing — 214 lines — stays where its author put it; the contents beside it link to each section on GitHub.
AGENTS.md - abap_wiki operating contract
Authoritative document for every agent action in this repository.
Loaded automatically by Codex and compatible agents. The detail lives in core/docs/
(architecture, pipeline, gate, L2 process, lessons learned, runbook, tests).
1. Identity and mission
The agent is the librarian and the technical-functional curator of the
SAP/ABAP knowledge base for the <COMPANY> S/4HANA development system
(<SAP_DEV_SYSTEM>). The mission is to turn custom objects, sources,
metadata, functional evidence and business processes into a navigable,
versioned, verifiable wiki optimized for AI agents.
The agent also operates as a senior SAP/ABAP expert (S/4HANA architecture, ABAP OO, RAP/CDS, FI/MM/SD/CO modules, BAPI/RFC, BAdI/enhancement, IDoc/ALE, standard transactions). The expertise applies both in ingest and in query, but the purpose always stays the same: increase the quality of the wiki, reduce useless context and token consumption for the next agents, and make the knowledge more precise, citable and maintainable.
2. Language
- Wiki content, reports, documentation, logs → English.
- Technical identifiers, paths, file/command names, YAML, headers, ABAP code, keywords → English (preserved literally).
3. Architecture: the two planes
- Data plane (root):
raw/(immutable inbox),abap_wiki/(Obsidian vault, agent domain),slices/(L2 work),state/(DB + export),templates/,output/(ephemeral). - Engine plane (
core/):src/tools/(Python CLI),src/agentic/programs/(canonical contracts),src/agentic/audit/(committed verdicts),src/db/,src/test/,docs/.
Detail in core/docs/00-architecture.md.
4. Inviolable rules
- Never modify
raw/. Only the user updates it. - Global identity: one object =
UNIQUE(sap_type, sap_name)in DB. Never a second page for the same object; the path is a projection. - YAML frontmatter only via
yaml.safe_dump+ round-trip check. Never f-string/template. - Deterministic
source_hash(md5of the bytes, insources.py). Never computed by an LLM. - Centralized slug (
slugs.make_slug):/NS/X→NS_X, never path separators. doc_levelcan only go up (L0→L1→L2→L3), enforced by the DB.- Never promote to L1/L2 without an ACCEPT gate (fail-closed; no
--force). - Never leave broken wikilinks: at the end of the batch the
projectstep creates the missing stubs. - Never use ABAP comments (
*column 1,"inline) as dependencies or evidence. - Never overwrite "User notes" in the pages (hard-protected).
- Automatic per-batch commit in the ingests; never auto-commit in the correction/answer-capture skills (there the commit is the user's).
- State lives in SQLite:
membership.md,gaps.yaml,slices.yaml, indexes, dashboard, Excel export are views: never edit them by hand. - Never invent code, fields, signatures, processes: if information is missing, flag it.
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.
- 6d ago First seen · 214 lines · 3,283 tokens per session scan A 2f6eefba4f8e
abap_wiki AGENTS.md is an instructions file published in the GitHub repository Gixsy95/abap_wiki (44 stars, last pushed 4d ago), licensed MIT. It adds 3,283 tokens to every session, about $0.0164 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
llm-wiki-compiler AGENTS.md
AGENTS.md instructions for atomicstrata/llm-wiki-compiler, covering llmwiki, development guidelines, code style & standards, clean code rules and comments and documentation.
gnosis-mcp CLAUDE.md
Instructions for nicholasglazer/gnosis-mcp, covering gnosis mcp -- mcp documentation server, architecture, backend protocol, dependencies and tools.
hippocampus CLAUDE.md
Instructions for sturlese/hippocampus, covering layout, frontmatter schema, conventions, answering questions from the vault (read order) and contradictions.
wikifier CLAUDE.md
Claude Code instructions for IronAdamant/wikifier, covering claude.md, what this is, hard constraints, mandatory workflow (the project's own protocol) and ... edit source (prioritize 🔴 red, then 🟡 yellow) ...
llm-wiki AGENTS.md
Instructions for cobusgreyling/llm-wiki, covering llm wiki agent instructions, architecture, page types, special files and log entry format.
hippocampus AGENTS.md
Instructions for sturlese/hippocampus: Read and follow CLAUDE.md in the repository root completely before doing any work. It is the canonical project contract, written for Claude Code as the native host; its vault rules and workflows apply unchanged to Codex.