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/localplugins/plugins/understandgit clone --depth 1 https://github.com/localplugins/pluginsWhat 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 | $0.00023 | $0.00397 |
| Opus 5 | $0.00012 | $0.00198 |
| Sonnet 5 | $0.00005 | $0.00079 |
| Haiku 4.5 | $0.00002 | $0.00040 |
Grade A, and why
understand 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
Understand
Make sense of a financial export. Arguments: $ARGUMENTS
Workflow
- Config. Ensure
money/categories.jsonexists; if not, run the/money-setupflow first. - Analyze. Delegate to the
analystsubagent: detect the column mapping, then run the toolkit (parse_report→categorize→aggregate→anomalies) — all figures fromDecimalmath, never estimated. - Surface skipped rows.
parse_reportreturns the clean transactions plus.skipped— rows that couldn't be parsed (malformed amount/date) or were ambiguous (both debit and credit populated). Flag these for the user; never drop or fabricate them silently. - Surface uncategorized transactions and propose new rules; never guess a category.
- Verify. Delegate the draft to the
figures-guardiansubagent — it checks the invariants and that every number traces to source rows; if it rejects, redo and re-check. - Write to
money/output/<name>/:categorized.csv— every transaction with its categoryreport.md— plain-English summary (income, spend by category, net cash flow, recurring)anomalies.md— flagged duplicates / outliers / recurring changesskipped.csv— rows that couldn't be parsed (malformed/ambiguous), for the user to fix — written only when there are any
- Summarize the highlights and list any skipped rows and uncategorized items for the user to resolve.
Never connect to the network or the user's accounts. Never invent a number.
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 · 24 lines · 23 tokens per session scan A d0b9d7699ff3
understand is a command published in the GitHub repository localplugins/plugins (5 stars, last pushed 1mo ago), licensed MIT. It adds 23 tokens to every session and 397 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
security-audit-static
Static security audit of AI-built code — map trust boundaries, cross-reference documented intent, self-refute every finding, and report only evidence-backed risks.
performance-audit-static
Static performance audit of AI-built code — find N+1 queries and request waterfalls, over-fetching, missing indexes, and caching opportunities, ranked by effort and impact.
sprint
Sprint lifecycle — plan a sprint, run a retrospective, or generate release notes.
document-app
Reverse-engineer an AI-built codebase into the system documents reviewers and auditors need — a core set (architecture, flows, permissions, variables) plus conditional docs (emails, cron, SEO, automation) when they apply.
analyze-test
Analyze A/B test results — statistical significance, sample size validation, and ship/extend/stop recommendations.
plan-okrs
Brainstorm team-level OKRs aligned with company objectives — qualitative objectives with measurable key results.