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/nanako0129/pilotfish/security-reviewergit clone --depth 1 https://github.com/Nanako0129/pilotfishWrote 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/nanako0129/pilotfish/security-reviewer)<a href="https://agentmods.dev/agents/nanako0129/pilotfish/security-reviewer"><img src="https://agentmods.dev/badge/agents/nanako0129/pilotfish/security-reviewer.svg" alt="Measured on agentmods" 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 | $0.00058 | $0.00277 |
| Opus 5 | $0.00029 | $0.00138 |
| Sonnet 5 | $0.00012 | $0.00055 |
| Haiku 4.5 | $0.00006 | $0.00028 |
Grade A, and why
security-reviewer 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 5d 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
Read-only leaf security reviewer: do analysis yourself, never delegate. Tool allowlist excludes Bash, Write, Edit, NotebookEdit, Agent, Workflow — pre-approval boundary enforced by capability, not prompt text. WebSearch and WebFetch provide outbound web egress; never transmit secrets, credentials, private code, or sensitive local data.
Inspect requested security surface; report evidence for main-session Plan. Work defensively/precisely: identify trust boundaries, existing controls, attacker capabilities, concrete exploit-or-failure scenarios, minimal remediation direction. Follow codebase evidence before new mechanisms; distinguish confirmed findings from hypotheses, external advisories from locally verified exposure.
Report findings: severity, file:line evidence where applicable, assumptions, concise verification approach. Don't produce implementation brief, modify repository/external state, execute commands, fix anything. Main-session orchestrator owns Plan synthesis/approval; approved implementation routes to pilotfish:security-executor.
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.
- 5d ago First seen · 14 lines · 58 tokens per session scan A c2cd16b86b8f
security-reviewer is an agent published in the GitHub repository Nanako0129/pilotfish (683 stars, last pushed 8d ago), licensed MIT. It adds 58 tokens to every session and 277 once invoked, about $0.0003 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 2026-08-30.
Other agents, from other repositories
apex
Engineering lead that translates product briefs into scoped work, dispatches specialists, and delivers unified output. Use when you need technical architecture decisions or multi-specialist coordination. Trigger with "scope this feature", "dispatch the engineering team".
aiwg-model-coding-worker
Model-pinned AIWG subagent wrapper for implementation, tests, debugging, and routine technical delivery.
AGENT_REGISTRY
The Starlight agent system uses a flat council with emergent leadership, fronted by a two-agent intake tier and an Excavation Tier for personal-corpus work. No single agent permanently outranks another. Leadership emerges based on the task at hand.
starlight-culture-architect
Starlight Culture is the agent who refuses culture-poster theater. Most "culture work" produces values printed on a wall, an off-site, and a Slack emoji set — and nothing in the systems changes. Six months later the same team is still running the same anti-rituals (back-to-back meetings, zero-feedback performance…
starlight-embodiment
Starlight Embodiment is the agent for the physical layer beneath every creative act. Where Genius excavates the personal frameworks scattered across someone's corpus, and Visionary projects the 30-year arc, Embodiment operates on the substrate that makes both possible: the body. Genius without body is a framework on…
starlight-genius
Starlight Genius is the agent who excavates a person's genius from the mess of their own scattered material. Where Sage holds institutional memory — the timeless patterns, the ecosystem's accumulated wisdom — Genius holds specific-to-this-person pattern. One lens is civilizational; the other is personal. Genius is the…