agentleFS
mtarcuremtarcure/claude-vibe-squad/.agents/skills/findings-filter/SKILL.md

findings-filter

Skill220 starsChanged 20 days agoMITView on GitHub

Use when you are sitting on a pile of candidate findings and must cut it down to what is real, reachable, and worth reporting — run each through reproduce-from-clean, reachability, attacker-gain, precondition, and deduplicate-by-root-cause gates, then record which gate every dropped candidate failed.

The file

We split this skill into the rules it gives an agent

2 rules. About 0 are common, 2 are distinctive, and 0 are very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈2

    Distinctive · estimate

    This author's own choices. Worth a look.

  • 0

    Very specific

    Tied to their project's names, files or tools. Probably skip.

  • Common
  • Distinctive
  • Very specific
  1. Distinctive:

    Check each survivor against the intended audience's bar — a program's scope, an operator's threat model, a reviewer's severity floor — and …

    line 18
  2. Distinctive:

    Every retained finding was reproduced, with the reproduction recorded.

    line 22

≈ “Common” is an estimate: at least 1 in 5 of the 372,429 skill files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.

Keep what fits and drop the rest. Your version, ready for your agents.
Files
SKILL.md

Findings Filter

Filter a candidate finding set down to what is real, reachable, and worth someone's attention — before it reaches a report.

Steps

  1. Reproduce first. A finding that has not been reproduced from a clean state is a hypothesis; label it so or drop it.
  2. Apply the reachability gate: name the entry point, the actor, the privilege level, and the concrete input that reaches the defect. No path, no finding.
  3. Apply the impact gate: state what an attacker gains that they did not already have. Findings whose only outcome is disclosure of already-public information, or self-inflicted harm by a privileged user, do not survive.
  4. Apply the precondition gate: list every assumption the finding needs. If the preconditions are individually unlikely and jointly required, say so and rank accordingly.
  5. Deduplicate by root cause, not by symptom. Ten call sites of one unsafe helper are one finding with ten instances.
  6. Separate defects from hardening suggestions. Hardening is legitimate output but must not be presented at defect severity.
  7. Check each survivor against the intended audience's bar — a program's scope, an operator's threat model, a reviewer's severity floor — and drop what is out of scope rather than padding.
  8. For every dropped finding, record the gate it failed. The dropped list is evidence of thoroughness and prevents the same candidate being re-raised next pass.

Acceptance

  • Every retained finding was reproduced, with the reproduction recorded.
  • Each retained finding names entry point, actor, privilege, and reaching input.
  • Each states the attacker's concrete gain, not just the anomaly.
  • Findings are deduplicated by root cause, with instances listed underneath.
  • Hardening is separated from defects, and every drop records which gate it failed.

Download now (ZIP)

Does it work?

Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.

No reviews yet

People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.

  • It worked
  • Partly
  • It didn't
What did you use it for? e.g. Reviewing pull requests in a Rails app

Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.

Your trail

Files you open will collect here, so you can find them again.

Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.

Now do this for your own.

Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.