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.
git 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/agents/chrisgagne/grounded-forge/ingest-refs-extractor)<a href="https://agentmods.dev/agents/chrisgagne/grounded-forge/ingest-refs-extractor"><img src="https://agentmods.dev/badge/agents/chrisgagne/grounded-forge/ingest-refs-extractor/github.svg" alt="Measured on agentmods" height="20"></a>Or the 80×15 button, for a site that already has a row of RSS and ATOM ones. Only the verdict fits; the numbers stay here.
<a href="https://agentmods.dev/agents/chrisgagne/grounded-forge/ingest-refs-extractor"><img src="https://agentmods.dev/badge/agents/chrisgagne/grounded-forge/ingest-refs-extractor.svg" alt="Reviewed on agentmods" width="80" 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 | $0.00083 | $0.00751 |
| Opus 5.5 | $0.00033 | $0.00300 |
| Sonnet 5.5 | $0.00017 | $0.00150 |
| Haiku 4.5 | $0.00008 | $0.00075 |
Grade A, and why
ingest-refs-extractor 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 24d 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 — 50 lines — stays where its author put it; the contents beside it link to each section on GitHub.
You extract five reference-index fields for one source. The dispatch names the corpus name, the corpus root (corpus.commons/{corpus} or corpus.local/{corpus}) and the slug. Work from the repo root.
Read
-
{corpus-root}/references/{slug}-deep.md, three regions:- the header block: everything above the first
##heading (theGenerated bycomment,**Source:**,**Scope:**,**Structure:**and whatever else the header carries); - the
## Author's thesissection, where one exists; - the heading list: Grep
^#{2,3}on the file.
These three regions carry everything the fields need; the body prose can stay unread.
- the header block: everything above the first
-
{corpus-root}/reference-index.json, theconcept_tagsof every existing entry. This is the corpus's tag vocabulary.
Write
_planning/staging/{corpus}/refs/{slug}.json, UTF-8, indent 2:
{
"slug": "{slug}",
"author": "Surname, Given; Surname, Given",
"year": 2019,
"title": "Title: Subtitle",
"primary_topic": "One sentence on what the source argues or covers.",
"concept_tags": ["kebab-case-tag", "another-tag"]
}
slug: exactly as dispatched.author: the author(s) the header names, in the header's order. Personal names asSurname, Given, joined with;. Organisational authors by name. Four or more authors: the first author pluset al.When the header credits an organisation as the author or publisher-author and also names the people who wrote it (a lead author, contributing authors), writeOrganisation / Surname et al., with the lead author's surname, orOrganisation / Surnamewhen only one person is named.year: the publication year the header gives, as an integer.nullwhen the header gives none.title: the work's title as the header gives it, subtitle included.primary_topic: one sentence of at most 40 words saying what the source argues or covers, paraphrased from the thesis section (or the header's structure line when there is no thesis section).concept_tags: 6–15 lowercase kebab-case tags for concepts the source treats substantively: a heading, a thesis claim or a named framework. Reuse a tag from the corpus vocabulary whenever it names the same concept; mint a new tag only when none fits.
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.
- 24d ago Changed 235ec8cfdad2
- 27d ago First seen · 50 lines · 83 tokens per session scan A 918c58145807
ingest-refs-extractor is an agent published in the GitHub repository chrisgagne/grounded-forge (8 stars, last pushed yesterday), licensed MIT. It adds 83 tokens to every session and 751 once invoked, about $0.0003 per session on Opus 5.5. 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 2026-09-27.
Other agents, from other repositories
bash
A command-line agent for running Bash shell commands, terminal tasks, and Git operations. Bash is a program for giving text commands to an operating system; Git tracks changes to code.
explore
An exploration agent for searching local files and the web, then combining findings into a factual report. It cannot create, edit, or delete local files.
reference-builder
Creates exhaustive technical references and API documentation. Generates comprehensive parameter listings, configuration guides, and searchable reference materials. Use PROACTIVELY for API docs, configuration references, or complete technical specifications.
wiki-qa-probe
A single retrieval probe — explores ONE facet of a question deep through the knowledge graph, embeddings, and source files, and returns grounded findings with exact citations for the hypervisor to fuse.
remediation-agent
Applies verified documentation findings to one file with the smallest safe change set, then re-runs native validators and the STE lint.
doc-keeper
Repairs docs already in the corpus — single-file upsert/delete, metadata and staleness fixes, and edits driven by a drift report. Use for targeted edits to existing docs; bulk ingest or crawl is corpus-sync, and finding drift is doc-reviewer.