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 commands/armanfatemi/nullius/groundgit clone --depth 1 https://github.com/armanfatemi/nulliusWhat 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 | $0.00016 | $0.00249 |
| Opus 5 | $0.00008 | $0.00125 |
| Sonnet 5 | $0.00003 | $0.00050 |
| Haiku 4.5 | $0.00002 | $0.00025 |
Grade A, and why
ground 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 yesterday.
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.
What it actually says
Verify Evidence Anchors with the nullius checker.
If $ARGUMENTS names one or more files or globs, run:
npx -y @nullius-inverba/claims check $ARGUMENTS
If no argument was given, locate the most recently modified .md file in the
plans directory (.claude/plans or ~/.claude/plans, newest first) and check
that file instead.
Then:
- If every marker verifies, say so in one line and stop.
- For each failing verdict, open the cited file, fix the citation in the
document — or, if the claim itself cannot be supported, move it to an
## Open questionssection. AFABRICATEDorCOUNT-MISMATCHverdict is not a citation typo: re-examine the decision that claim was supporting, and say so explicitly. - If the document contains load-bearing claims about existing code with no anchors at all, flag them — a document with no grounding markers is not a pass.
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.
- yesterday First seen · 28 lines · 16 tokens per session scan A 047badfe8e27
ground is a command published in the GitHub repository armanfatemi/nullius (4 stars, last pushed yesterday), licensed MIT. It adds 16 tokens to every session and 249 once invoked, about $0.0001 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 commands, from other repositories
drift
Post-implementation spec drift check — verify the implementation matches existing OpenSpec specifications.
brainblast-deploy-plan
Deployment Intelligence for Anchor programs. Run before anchor deploy so you know how much SOL you need and the exact ordered transaction sequence — instead of working it out by hand.
brainblast-oracle
Live On-Chain Intelligence: is the oracle fresh? Before you price anything against a Pyth / Switchboard / Chainlink feed, check when its account was last written.
brainblast-rico-maps
Runs a two-part token check on any Solana contract address (CA).
incident
Coordinates the live response to an active production incident — timeline, status, and resolution — then hands off to /renata:retro for the post-mortem.
landscape
Post-PRD competitive research — maps similar solutions (anchored on the PRD), then co-curates the result with you to surface differentiation gaps that become candidate features. Sources mandatory.