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/emergent-wisdom/sema/agents-mdgit clone --depth 1 https://github.com/emergent-wisdom/semaWrote 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/emergent-wisdom/sema/agents-md)<a href="https://agentmods.dev/instructions/emergent-wisdom/sema/agents-md"><img src="https://agentmods.dev/badge/instructions/emergent-wisdom/sema/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 | $0.07325 | $0.07325 |
| Opus 5 | $0.03662 | $0.03662 |
| Sonnet 5 | $0.01465 | $0.01465 |
| Haiku 4.5 | $0.00732 | $0.00732 |
Grade A, and why
sema 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 4d 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 — 508 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Repository Agent Guide
This file is the repository-level operating context for coding agents. Detailed domain rules remain in the linked specifications; do not duplicate or invent them from prompt context.
Development Workflow
- Work on a focused feature branch and use a pull request. Do not merge until required CI is green.
- Keep unrelated user changes intact and surface unrelated defects separately.
- Use
Henrik Westerberg <[email protected]>for locally authored commits. GitHub-created merge commits may use the account's noreply address; merge locally when every commit must use the organization address. - Website and frontend experiments must remain on an isolated branch and staging service until Henrik explicitly approves production deployment.
- Paper typography or stylesheet work must remain separate from semantic paper updates unless visual redesign is the stated scope.
- Any change that touches
paper/must be built withscripts/compile_paper.sh, never barepdflatex. The script regenerates the tables, the pattern-card appendix, the prose hash references and the stats macros, then runs bibtex and three pdflatex passes. Bare pdflatex leaves those stale: it once shipped 49 outdated hash stubs in the prose and a wrong pattern count, and a stale generateddocs/information/audit.mdfrom the same class of mistake failed CI on the 3.12 job, which is the only job that runs the vocabulary-workflow check.
Vocabulary Changes
Read these before editing patterns:
docs/guides/authoring.mddocs/guides/lifecycle.mddocs/specification/schema.mddocs/specification/validation.md
The database is authoritative. Copy exported JSON from data/vocabulary/ into
data/staging/, edit staging, update data/design_critique.json, preview the
manual, then apply. Never edit canonical vocabulary exports directly.
To apply, prefer:
python scripts/apply_vocabulary_change.py
It runs the four steps in the only order that works and stops at the first
failure. It ignores any database selected by sema use and pins every
subprocess to this checkout's authoritative data/taxonomy.db.
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.
- 4d ago First seen · 508 lines · 7,325 tokens per session scan A 6ce77a8813e9
sema AGENTS.md is an instructions file published in the GitHub repository emergent-wisdom/sema (13 stars, last pushed 3d ago), licensed MIT. It adds 7,325 tokens to every session, about $0.0366 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
dibs AGENTS.md
AGENTS.md instructions for Agenxy/dibs, covering agents.md: orientation for agents working on dibs, what this repo is, layout, rules you must not break and working here.
tiding AGENTS.md
Instructions for iXanadu/tiding, covering tiding, sources of truth, the backlog — read every session, sweep every wrapup, project structure and commands.
building-multiagent-systems CLAUDE.md
Instructions for 2389-research/building-multiagent-systems, covering building multi-agent systems plugin, skill included, building-multiagent-systems, patterns and six coordination patterns.
dibs CLAUDE.md
Claude Code instructions for Agenxy/dibs, a project described as: Keeps your agents in the loop about each other. See what the fleet is doing, then message, sync and send files between agents. Dibs reports; it never acts.
shift-mcp-server AGENTS.md
AGENTS.md instructions for cyanheads/shift-mcp-server, covering developer protocol, what's next?, core rules, patterns and tool.
shift-mcp-server CLAUDE.md
Claude Code instructions for cyanheads/shift-mcp-server, covering developer protocol, what's next?, core rules, patterns and tool.