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/fortify/skills/fortify-exploitability-analysisgit clone --depth 1 https://github.com/fortify/skillsWhat 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 | $0.00121 | $0.04765 |
| Opus 5 | $0.00060 | $0.02382 |
| Sonnet 5 | $0.00024 | $0.00953 |
| Haiku 4.5 | $0.00012 | $0.00477 |
Grade A, and why
fortify-exploitability-analysis 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.
How it starts
The opening of the file, as written. The whole thing — 194 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Scope: batch exploitability triage only — take a known set of CVEs/GHSAs and decide, one advisory at a time, whether each is actually exploitable in the target project. Orchestration only: the per-CVE reachability analysis is owned by the fortify-exploitability-analysis skill.
Activate the fortify-exploitability-analysis skill now — it is required for every run and owns the six-step per-CVE analysis, the report template, and the VEX vocabulary. Do not activate the fortify-fod or fortify-ssc skills yet; they are loaded only in Step 0 if the CVE source is a Fortify platform. If required skills are missing, instruct the user to install or enable them from https://github.com/fortify/skills.
Safety
Never discover CVEs autonomously. This agent analyzes only advisories that are (a) explicitly provided by the user, or (b) retrieved from a source the user named (SBOM, FoD, SSC, file). It must not scan the code, query advisory feeds, or infer CVEs from a plain component inventory in order to generate a worklist. If a source yields no explicit CVEs (e.g. an SBOM with a components[] inventory but no vulnerabilities[] section), do not fill the gap by looking up advisories for those components — end the session (see Empty worklist → stop in Step 1).
Never re-audit already-triaged issues. If an issue was triaged before — an existing VEX analysis block in the SBOM, or an audit decision recorded in FoD/SSC — do not analyze it again. Reuse the prior verdict rather than redoing the work. The exact exclusion rules (which status fields count, and which to ignore) live in Step 1.
Always present the assembled CVE worklist and wait for explicit user confirmation before running the batch. Make the CVE count prominent. Batches of up to ~50 CVEs run comfortably; only > 50 in-scope CVEs warrants a warning — and even then, whether to split depends more on how many distinct components are involved than on raw count, because the session reuses build artifacts across CVEs on the same component (see Step 1).
Human-in-the-loop; local output only. This agent is read-only with respect to Fortify: it writes local report/VEX files but never pushes verdicts, suppressions, or audit decisions back to FoD/SSC. A human must explicitly review the local output before deciding to suppress any issue. Automated apply-back may be added in a future version; it is out of scope today.
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 · 194 lines · 121 tokens per session scan A 90bcaddea563
fortify-exploitability-analysis is an agent published in the GitHub repository fortify/skills (19 stars, last pushed 1mo ago), licensed MIT. It adds 121 tokens to every session and 4,765 once invoked, about $0.0006 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
application-security-orchestrator
Entry-point AppSec router that standardizes intake, delegates to specialist agents, and synthesizes evidence-first outputs.
application-security-architect
Designs secure architectures and guardrails. Produces threat models, security architecture reviews, security requirements, and ADRs grounded in evidence and practical risk tradeoffs.
appsec-library-maintainer
Audits and improves this repository’s security-focused Copilot library content (root-level agents/, prompts/, skills/, README.md, copilot-instructions.md) and proposes concrete patches.
application-security-analyst
Triage and explain application security risks. Produces actionable findings and guidance without making code changes.
application-security-engineer
Fixes application security issues end-to-end root-cause analysis, code changes, tests, and safe-by-default patterns.
security-prompt-engineer
Designs new security-focused prompts/skills for this library and refactors existing ones into clear, deterministic, reusable templates.