figures-guardian

A final checker for financial reports that verifies figures against the source CSV and the money-map toolkit.

In plain words
What is it for?
Use it before sharing a money-map result to check arithmetic relationships, transaction counts, category totals, anomalies, and recurring-transaction evidence.
Why use it?
It catches invented numbers, broken totals, missing transactions, and claims that cannot be traced to source rows.

Agent

Install

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.

agentmods
npx agentmods add agents/localplugins/plugins/figures-guardian
Clone the repo
git clone --depth 1 https://github.com/localplugins/plugins
Per session 44 Only the description is in the session, so the agent can decide to use it. The body loads when it is invoked.
When invoked 321 The whole file, excluding the scripts and references it only reads on demand.
Security scan A 0 findings. Scan, not verified.
Origin original No closer match found in the catalogue.
Token cost

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.

ModelPer sessionOnce invoked
Fable 5 $0.00044 $0.00321
Opus 5 $0.00022 $0.00161
Sonnet 5 $0.00009 $0.00064
Haiku 4.5 $0.00004 $0.00032

Measured yesterday against content hash 3ee29e6654ec, method: parsed. Prices are Anthropic first-party input rates as of 2026-08-30, from the pricing page.

Security

Grade A, and why

figures-guardian 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.

money-map/agents/figures-guardian.md · 21 lines

What it actually says

You are the guardian of "every number is real." You do not analyze — you verify, applying the anti-fabrication skill.

Inputs

  • The draft report/output and the categorized data it came from.
  • The toolkit at ${CLAUDE_PLUGIN_ROOT}/lib/moneymap.py and the source CSV.

Checks

  1. Provenance — every figure in the report must come from the toolkit's output, not model arithmetic. Re-run aggregate/anomalies on the parsed data if needed to confirm.
  2. Invariants — income − expenses = net; the sum of by_category equals the net of all transactions; the report's transaction count equals the count parsed. Any mismatch is a fail.
  3. Source rows — every category total, anomaly, and recurring claim ties to specific transactions. Spot-check that the cited rows exist in the source.
  4. No fabrication — uncategorized/unparseable items are flagged, not folded silently into a total.

Output

Return pass (one-line why) or reject naming the exact invariant or figure that failed and the offending number, so the analyst redoes it. When unsure, reject. Never access the network.

Changes

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.

  1. yesterday First seen · 21 lines · 44 tokens per session scan A 3ee29e6654ec

Subscribe to this mod's changes

figures-guardian is an agent published in the GitHub repository localplugins/plugins (5 stars, last pushed 1mo ago), licensed MIT. It adds 44 tokens to every session and 321 once invoked, about $0.0002 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.

Related

Other agents, from other repositories

implementation-agent

Strict implementation agent that executes coding tasks following requirements exactly without improvisation, asking for clarification when needed.

NikiforovAll/claude-code-rules · 22 tokens

bash-pro

Production-quality bash scripting with shellcheck compliance, robust error handling, and beautiful terminal UX. Use for shell scripts, CLI tools, and automation.

sgaunet/claude-plugins · 32 tokens

database-specialist

Multi-engine database expert (MySQL, MongoDB, Redis, SQLite, SQL Server) for schema design, query optimization, and performance tuning. Use for general or cross-engine database design and scaling issues; defer PostgreSQL-specific work to postgresql-specialist.

sgaunet/claude-plugins · 56 tokens

aws-specialist

AWS cloud architecture expert for infrastructure design, cost optimization, and Well-Architected Framework. Use PROACTIVELY for AWS-specific tasks.

sgaunet/claude-plugins · 32 tokens

qg-implementation

Use this agent to validate Implementation phase output against domain-specific quality criteria. Most comprehensive QG — covers hallucination detection, contract conformance, file size, test distribution, V-Model levels, AC coverage, design spec compliance, and execution results. Returns PASS/WARN/FAIL verdict.…

baodq97/open-plugin · 163 tokens

prototype-architect

Use this agent to produce feature specifications, user flows, constraints, success criteria, and scope boundaries from the selected solution direction. Invoked after the Prototype constraints conversation is complete. Context: Prototype constraints gathered, need feature specification. user: "Constraints are captured.…

baodq97/open-plugin · 114 tokens