Skip to content
SkillModel-led4,540 tokens in use

hunt-dispatch

The orchestration skill behind the /hunt command: fingerprints a target and loads the right red team or web app pentest skill set under a strict scope frame.

14,614 stars on the repository MIT Licenseelementalsouls/Claude-BugHunter · skills/hunt-dispatch
/plugin marketplace add elementalsouls/Claude-BugHunter /plugin install claude-bughunter@elementalsouls
View source

Install command as written in the repository's README

Skill label

measured from its files

What installing hunt-dispatch costs

77tokens idle
4,540tokens in use

Idle is what sits in context while the skill is installed. In use is its instructions, loaded when it runs. Counts are approximate, made offline with cl100k_base.

Model dependence Model-led

Script-ledMixedModel-led

Mostly instructions, so how well it works depends on the model that follows them.

This describes the skill's structure, how much of it is scripts. It does not say how well it works.

Touches

shelldetectednetworkdetectedwrites filesdetectedcredentialsdetected

Declared tags come from the skill's frontmatter; detected ones from its scripts, its code blocks and the commands it names in inline code. This is detected, not audited.

How it scores

Quality

How well the skill is written, counted from its own files: its frontmatter, its steps, examples, stated output and constraints, its size and whether every file it ships is one it names.How we score it →
Excellent
  • Frontmatter12/15
  • Structure25/25
  • Examples12/15
  • Output15/15
  • Size15/15
  • Resources10/15

Security

Starts at 100 and loses points for each thing in the skill's files that a reader should look at first.How we score it →
Good
  • Runs shell commands−5

    does this on the machine, and the frontmatter does not say so

    SKILL.md, line 46
  • Reaches the network−10

    does this on the machine, and the frontmatter does not say so

    SKILL.md, line 50
  • Writes files−8

    does this on the machine, and the frontmatter does not say so

    SKILL.md, line 96
  • Reads credentials−15

    does this on the machine, and the frontmatter does not say so

    SKILL.md, line 131

Repository health

Pre-computed score based on development activity, maintenance, community, maturity, and trend momentum.How we score it →
Excellent
  • Development activity100
  • Maintenance84
  • Community68
  • Maturity28
  • Momentum40

hunt-dispatch is a skill for Claude that loads the correct skill set once a /hunt engagement has an authorized scope and a mode answer (red team or web application pentest, blackbox or greybox). It runs as the choke point every /hunt session passes through before any individual hunt-* skill loads, and the /hunt command itself is the only intended caller.

The skill opens with a mandatory soft-404 baseline that every host must clear before any path is probed, because many modern front ends return HTTP 200 for paths that do not exist and a status code alone proves nothing. It then fingerprints live hosts, maps the signals it finds to the matching platform or hunt-* skills under a tiered load budget, prints a deterministic taxonomy of what was loaded, and states strict scope-inheritance rules for any work handed to a subagent. It is for engagements where written authorization for the named scope has already been established.

What You Get

  • A mandatory soft-404 baseline step that records status, byte length and body hash per host before any path is treated as a real finding
  • A fingerprint-to-skill mapping table covering identity providers, enterprise appliances, cloud markers, and web framework signals
  • A tiered load-budget system that caps platform-skill loading at eight and defers any further matches to a named list for later use
  • Fixed skill-set menus for mode=redteam and mode=wapt, each ending in the matching report-format skill
  • An authenticated-testing preflight that validates greybox credentials before fanning out across the authenticated hunt-* set, and explicit subagent scope-inheritance rules including a deny-before-allow verb list

Common Use Cases

  • Fingerprinting a set of live hosts in a /hunt engagement and determining which enterprise platform or framework skills to load
  • Establishing a soft-404 baseline for a CDN-fronted or SPA target before treating any discovered path as a real finding
  • Passing explicit, data-formatted scope boundaries and a deny-before-allow verb list to a subagent before delegating part of an authorized engagement

What Makes It Unique

The skill functions purely as orchestration infrastructure for the bundle, routing to the hunting skills that do the actual class-specific work. Its subagent-scope section carries lessons from a real engagement where a subagent told to stay read-only still created a live production record and wrote to an out-of-scope bucket: scope passed as data, a deny list of action verbs checked before any allow list, and discovered hosts treated as out of scope until the operator says otherwise.

Under The Hood

The skill is a single SKILL.md of guidance with no bundled scripts. Following it, the agent sends HTTP requests to establish a soft-404 baseline and to fingerprint live hosts, reading response headers and body content for platform and framework markers, then invokes other skills by name through the Skill tool based on what it found. It reads from and may create files under a recon/ working directory for per-host results. Use it only within an engagement scope that has already been authorized.

Join founders building with open source

Opinionated takes, migration guides, cost-saving tips, and insights from the open source ecosystem.

Subscribe on SubstackJoin 750+ subscribers