Agentry

blaktron/vulnerability-applicability-check

autonomousv0.2.0

Checks whether one named vulnerability is present, loaded and reachable in a codebase, judged on the code alone with no external…

Declared capabilities

network: 2 hostsdata: codeknowledge: ships 1

Ships knowledge: sast_skills (809.6 KB), in the release archive under knowledge/<name>/, scanned with the version. What the scan read

Capabilities are the author's declaration and the basis for policy and scanning. The registry cannot enforce them at run time. Your runtime should.

MCP servers & tools

agentry run serves exactly the tools this contract declares, through a per-run MCP gateway. An undeclared capability is a tool that does not exist at run time.

Tools the runner grants

Manifest declarationTool(s) granted
(always)write_report — reports → the run's output folder
(always)list_inputs, read_input, search_inputs — operator-provided input files, read-only
(always)scratch_list, scratch_read, scratch_write, scratch_search — the run's own scratch folder at /scratch, private to the run and deleted at its end; the runner's grant, not a line of the contract
(always)export_pdf — a Markdown report as a PDF in reports/, laid out in your report design; the runner's grant, not a line of the contract
(always)grep, glob — search file contents by regular expression and list files by path pattern (ripgrep), one area a call: the inputs, attached knowledge, /scratch; the runner's grant, not a line of the contract
network.egressfetch — pinned to api.osv.dev, services.nvd.nist.gov

MCP servers involved

  • QuickAgent Tools — the runner's companion server, served through a per-run Docker MCP Gateway; its tool surface is exactly the table above.
  • Operator-cataloged MCP servers may be attached for the declared dataAccess category (code); undeclared categories are never wired in.

In an embedded install

Installed into an existing coding agent (agentry install), the same contract maps to the host's built-in tools — enforcement there is the host's. Allowed tools: WebFetch(domain:api.osv.dev), WebFetch(domain:services.nvd.nist.gov).

Install

agentry install blaktron/vulnerability-applicability-check@0.2.0
git clone https://secureagentry.net/blaktron/vulnerability-applicability-check.git
git clone ssh://git@git.secureagentry.net/blaktron/vulnerability-applicability-check.git

claude-skill is this agent's default target, so the flag is not needed.

README.md

vulnerability-applicability-check

Checks whether a package's known vulnerabilities (all of them, in one run) or one named flaw are present, loaded and reachable in a codebase, judged on the code alone with no credit for external mitigations.

How to ask

  • By package (recommended): PyPI urllib3 2.5.0, or with IDs: PyPI pillow 12.3.0: GHSA-…, CVE-…. Several packages: separate with semicolons. Without IDs, it asks OSV for every advisory on that exact version.
  • From a reachability audit: supply findings.json from blaktron/vuln-reachability-audit 0.12.0+ as an input file and answer from findings.json (optionally priority high).
  • Single vulnerability: a CVE or GHSA id, or a class such as SSRF.

What you get

  • reports/vulnerability-applicability.md: per package, presence, how the code uses it, a verdict per advisory, a package verdict and one upgrade target.
  • reports/applicability.json: the same verdicts in the vulnerability-applicability/1 schema.

Changes in 0.2.0

  • Package mode: one run covers every advisory for a package, and several packages per run, instead of one run per CVE.
  • Findings mode: reads findings.json from the reachability audit.
  • The codebase map and each package's usage map are built once and reused across advisories; each advisory is still judged on its own trigger.
  • Alias deduplication and an explicit version-range check per advisory.
  • A work list tracked to the end, so long lists are not cut short.
  • JSON output alongside the Markdown.
  • Uses the shipped sast_skills knowledge pack (0.1.0 said none shipped).
  • No new network hosts; NVD is used only when OSV lacks detail.

Agentry uses only strictly necessary cookies — they sign you in and keep the site secure. Nothing here tracks you, and there are no advertising or analytics cookies. Privacy policy