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/chrisgagne/grounded-forge/claude-mdgit clone --depth 1 https://github.com/chrisgagne/grounded-forgeWrote 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/chrisgagne/grounded-forge/claude-md)<a href="https://agentmods.dev/instructions/chrisgagne/grounded-forge/claude-md"><img src="https://agentmods.dev/badge/instructions/chrisgagne/grounded-forge/claude-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 | $0.03658 | $0.03658 |
| Opus 5 | $0.01829 | $0.01829 |
| Sonnet 5 | $0.00732 | $0.00732 |
| Haiku 4.5 | $0.00366 | $0.00366 |
Grade A, and why
grounded-forge 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 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.
How it starts
The opening of the file, as written. The whole thing — 143 lines — stays where its author put it; the contents beside it link to each section on GitHub.
grounded-forge, system instructions
You are working on the grounded-forge repo. The distributions this repo compiles ship their own CLAUDE.md, sourced from per-corpus build profiles under corpus.commons/{corpus}/build-profiles/ or corpus.local/{corpus}/build-profiles/.
What this repo is
A working example of the reference × task matrix architecture: a retrieval pattern that pre-projects sources to task domains at ingestion time so runtime is lookup, not reshape. Task-axis projection moves from generation time to ingestion time, produced by an LLM operating under a structured protocol that enforces source-only citation discipline.
Lenses are per-distillation modifiers applied where they materially reweight what's salient, not a third axis. Specs live at corpus.commons/{corpus}/lenses/ and are designed via the creating-lenses skill at .claude/skills/creating-lenses/SKILL.md; per-distillation applicability is decided at Pass G of creating-distillations. Each lens carries a visibility: frontmatter field (open, open-nc, copyrighted, confidential, or personal) parallel to references' Scope:; the build's max_visibility: ceiling per profile filters which lenses ship.
For the architectural argument, read in this order:
docs/reference/vocabulary.md: the noun/verb grammar. Read first; the rest assumes it.docs/architecture/overview.mddocs/architecture/llm-epistemology.md: why the matrix exists. The Larman's-Law-4 / LLM-epistemology problem the architecture is solving against.docs/architecture/projection-time.mddocs/architecture/two-layer-indexes.mddocs/architecture/matrix-pattern.mddocs/architecture/source-integrity.mddocs/architecture/known-architectural-limits.md: three limits intrinsic to the design.docs/architecture/capability-binding.md
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 · 143 lines · 3,658 tokens per session scan A 2c9b1acdb652
grounded-forge CLAUDE.md is an instructions file published in the GitHub repository chrisgagne/grounded-forge (7 stars, last pushed 15d ago), licensed MIT. It adds 3,658 tokens to every session, about $0.0183 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 instructions, from other repositories
LightRAG AGENTS.md
AGENTS.md instructions for HKUDS/LightRAG, covering repository guidelines, project overview, project structure, module layout (lightrag/) and core architecture.
manifold AGENTS.md
Instructions for intelligencedev/manifold, covering go/manifold, coding conventions, project structure & module organization, package organization and dependency injection.
LightRAG CLAUDE.md
Claude Code instructions for HKUDS/LightRAG, a project described as:
manifold CLAUDE.md
Instructions for intelligencedev/manifold, covering go/manifold, coding conventions, project structure & module organization, package organization and dependency injection.
ai-dial-core CLAUDE.md
Claude Code instructions for epam/ai-dial-core, covering claude.md, build & run, set credentials via environment variables, build (skip tests) and run all tests.
ai AGENTS.md
AGENTS.md instructions for vercel/ai, covering agents.md, project overview, repository structure, key directories and core package dependencies.