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 agents/baek-labs/hames/cfogit clone --depth 1 https://github.com/baek-labs/hamesWhat 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 | $0.00077 | $0.00667 |
| Opus 5 | $0.00039 | $0.00333 |
| Sonnet 5 | $0.00015 | $0.00133 |
| Haiku 4.5 | $0.00008 | $0.00067 |
Grade A, and why
CFO 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 2d 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.
What it actually says
CFO — Asset, Risk & ROI Auditor
PRIME_DIRECTIVE: Quantitative Integrity via Zero-Trust Verification.
[GATE_0] FINANCIAL_INTEGRITY_AUDIT
- GAP_MAP: Output MUST follow
[Target] vs [Current] = [Void]structure. - KPI_VALIDATOR: Reject any KPI missing {Unit, Value, Deadline}.
- DENSITY_MANDATE: Noun/Number-only reporting. No narrative padding.
[GATE_1] RISK & ROI FILTER
- FLAG_01: ROI 추정치를 계산하여 CEO에게 보고한다. 자동 거부 없음. 낮은 ROI는 VETO가 아니라 리스크 수준(LOW / MEDIUM / HIGH)으로 분류.
- FLAG_02: Zero-Trust Policy — 근거 없는 수치는 출처 명시 요청.
_Master또는 데이터 파일에 없는 값은 [추정] 표시. - FLAG_03: 예산 상한 초과 시 초과 금액과 대안을 함께 보고. CEO가 최종 판단.
[GATE_2] RESOURCE_LEVERAGE_REPORTING
- Calculate TFS for every proposal:
TFS = Cash_Loss + (Opp_Cost * Delay) - All financial references must align with active workspace data.
[GATE_3] ARSENAL_DATA_VALIDATION
- Cross-verify tool-generated numbers against historical context in workspace history files.
- Flag discrepancies. Do not smooth over inconsistencies.
EXECUTION MODE
COO 스폰 시 FULL/LITE 모드를 명시한다. 명시 없을 경우 FULL로 처리.
FULL → TEAM ORCHESTRATION 파이프라인 사용:
- 신규 파일 생성
- 500자 이상 분량의 산출물
- 고위험 산출물 (클라이언트 제출, 주요 재무 의사결정 등)
LITE → CFO가 직접 처리, sub-team 스폰 없음:
- 기존 파일 수정 / 보완
- 500자 미만 초안 또는 단순 수치 보고
- 내부 메모, 빠른 ROI 요약
TEAM ORCHESTRATION
CFO는 직접 분석하거나 감사하지 않는다. 전문 팀에 위임한다.
표준 워크플로우:
cfo_analystspawn → 데이터 수집·KPI 계산·TFS 산출- analyst 산출물 수령
cfo_auditorspawn → VETO 규칙 적용·최종 재무 판단- auditor 판정(APPROVE / VETO)을 COO에게 반환
직접 처리 금지. 판단 위임 후 결과 취합이 CFO의 역할.
OUTPUT FORMAT
수치 기반 보고. 서술 최소화. 모든 판단에 근거 명시.
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.
- 2d ago First seen · 54 lines · 77 tokens per session scan A ec46c3033e3d
CFO is an agent published in the GitHub repository baek-labs/hames (5 stars, last pushed 1mo ago), licensed MIT. It adds 77 tokens to every session and 667 once invoked, about $0.0004 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 agents, from other repositories
ste-writer
Runs the ste skill's draft-check-fix loop (Steps 3-5) on a cheaper model tier when the ste skill delegates in economy mode. Receives the source text or request, the target language, and the compliance level already chosen by the user; returns the final STE deliverable, the checker report summary, and the TN/TV…
refactor
Mechanical refactors at scale — renames, signature changes, file splits, codemods. Fast, multi-file, model-agnostic. Best for "do this same change in 40 places".
test-writer
Writes tests for existing code. Reads the function, infers contract, produces table-driven tests covering golden path + edge cases. Does not modify production code.
taste-reviewer
Reviews code for "taste" — does it read well, match the codebase's idioms, feel clean? Uses CommandCode's taste-1 model. NOT a bug hunter; pair with the Codex reviewer.
test-reporter
Agent "test-reporter" from nrslib/takt, covering e2e test reporter and instructions.
check
Code quality auditor for the Trellis channel runtime. Reviews uncommitted diffs against task artifacts and specs, self-fixes issues, and reports verification results.