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/provasign/prism/claude-mdgit clone --depth 1 https://github.com/provasign/prismWhat 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 | $0.00684 | $0.00684 |
| Opus 5 | $0.00342 | $0.00342 |
| Sonnet 5 | $0.00137 | $0.00137 |
| Haiku 4.5 | $0.00068 | $0.00068 |
Grade A, and why
prism CLAUDE.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.
This is a copy
88% identical to prism copilot-instructions.md — 14 lines differ, which has more behind it and is treated as the original. This page carries a canonical link to it rather than competing with it.
How it starts
The opening of the file, as written. The whole thing — 48 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Before you commit or release — run the regression suite
CI runs these on push (.github/workflows/ci.yml; engine quality via
provasign/research/.github/workflows/engine-invariants.yml). Run them locally
before tagging a release:
go test ./...— unit suite; must be green.- Engine ceiling regression (no LLM): from the research repo,
python3 harness/ci_invariants.py --prism ~/bin/prism— asserts change-impact recall/precision, missing-implementations==[], and index determinism against committed ground truth. A drop here is a real completeness regression.
Do NOT tag a release with either red.
Prism — context delivery
Prism indexes this repo's call and type graph. grep finds the same lines — prism tells you which of them matter (file-level precision 0.51 -> 0.91 on change tasks) and answers in one call where search-then-read takes several.
Route by the question. One call, and treat its result as final:
| Question | Call |
|---|---|
| where is X? | prism_search(query="X") — searches symbol names AND raw text. Several names? You MUST batch: query=["X","Y"] — one search per name wastes a turn each. Know where to look: path=, glob=, files_only=true |
| a literal string, message or config key | prism_search(query="...", scope="text") — pure grep, cheapest. Use it for TEXT; leave the default for code |
| EVERY site of X (rewrite them all, count them) | prism_search(query="X", exhaustive=true) — results are capped at 25 by default and a capped answer to a completeness question looks complete. Say exhaustive; add files_only=true to keep it cheap |
| X, plus the lines around it | prism_search(query="X", context=N) — one call instead of search-then-prism_read |
| read one function, or one file | prism_lookup(name="pkg.Func") / prism_read |
| give me the code for X, ready to edit | prism_query(task="<label>", terms=["X"]) — keys on terms; the wording changes nothing |
| who breaks if I change X? | prism_change_impact(query="Type.method") |
| is my diff complete? | prism_verify |
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 · 48 lines · 684 tokens per session scan A 2e22461adb1c
prism CLAUDE.md is an instructions file published in the GitHub repository provasign/prism (5 stars, last pushed 2d ago), licensed Apache-2.0. It adds 684 tokens to every session, about $0.0034 per session on Opus 5. A static security scan graded it A with 0 findings. It is 88% identical to prism copilot-instructions.md, differing in 14 lines, and is treated as a copy.
Other instructions, from other repositories
kartograf AGENTS.md
Instructions for dev-manul/kartograf, covering agents.md, what this is, build & test, layout and hard-won rules (do not relearn these the hard way).
ollama CLAUDE.md
Claude Code instructions for ollama/ollama: See AGENTS.md for the shared agent instructions for this repository.
azure-sdk-for-go go-code.instructions.md
Instructions for Azure/azure-sdk-for-go: All code should follow the guidelines from the Azure Go SDK Guidelines. This document is a summary of the most important guidelines to follow when contributing to the Azure Go SDK.
gorest AGENTS.md
Instructions for pilinux/gorest, covering agents.md, project overview, build and run commands, build and tidy dependencies.
chatgpt-cli CLAUDE.md
Instructions for kardolus/chatgpt-cli, covering chatgpt-cli — release runbook, prerequisites, 1. cut the release, 2. publish the github release + binaries and 3. update the homebrew tap.
Sighthound AGENTS.md
Instructions for Corgea/Sighthound, covering claude, commands, python / django test naming and agent skill.