feature-marker-e3b0c4 — independently scanned and version-tracked by SaferSkills.
SaferSkills independently audited feature-marker-e3b0c4 (Agent Skill) and scored it 82/100 (green). The audit ran 55 deterministic rules across Security, Supply Chain, Maintenance, Transparency, and Community; it found 2 high-severity and 0 lower-severity findings. The full rule-by-rule trace and per-finding evidence are below. Free, methodology-open.
Findings & checks · 2 flagged
The text {match} is the classic direct prompt-injection phrasing. Placed in a skill body that the agent reads as trusted instructions, it tries to make the agent abandon its prior rules and follow whatever comes next — a full system-prompt override.
ignore/disregard/forget … previous instructions sentence.Every scanned point with the score it earned and what moved between them.
First recorded scan — no prior version to compare against.
The primary manifest — the file an agent reads to learn what this artifact does.
Use this skill when the user asks to implement, plan, test, or hand off a feature with Feature Marker.
The normal invocation is a plain LLM prompt:
Use feature-marker to implement <feature-slug>.Run through the feature by default. Do not use an interactive menu, old CLI workflow commands, checkpoint JSON, or a JavaScript workflow engine. The package installer only installs this skill.
short kebab-case slug from the feature request.
any existing tasks/{slug}/prd.md, tasks/{slug}/techspec.md, and tasks/{slug}/tasks.md.
a worktree or cleanup path;
main, master,develop, or trunk, create feature-marker/{slug} unless the user gave a branch name. Use a git worktree only when the checkout is dirty or the user asks for one.
tasks/{slug}/ using the bundled templates:templates/prd-template.md -> tasks/{slug}/prd.mdtemplates/techspec-template.md -> tasks/{slug}/techspec.mdtemplates/tasks-template.md -> tasks/{slug}/tasks.mdReplace {slug} and {feature_title}, fill every relevant section, and do not leave unresolved template placeholders in committed artifacts.
missing acceptance criteria, weak task order, risky files, data/migration impact, missing tests, unclear edge cases, and handoff gaps.
tasks/{slug}/ before implementation. Ask theuser only when a finding changes scope or requires a product decision. Otherwise continue without a generic artifact approval gate.
tasks/{slug}/tasks.md. Keep changes scoped tothe feature and preserve unrelated local edits.
it is in scope; otherwise report the exact blocker and stop.
unless the user prohibited commits. Do not push or open a PR automatically.
git push -u origin <branch>
gh pr create --base <base-branch> --head <branch>The canonical state is user-facing markdown:
tasks/{slug}/
prd.md
techspec.md
tasks.mdOptional verification notes may also live under tasks/{slug}/ when they help future continuation, but checkpoint JSON is not the source of truth.
There are no CLI workflow modes. Treat these as prompt intents:
full: run the complete PRD -> TechSpec -> Tasks -> implementation grill ->implementation -> tests -> branch handoff flow.
tasks-only: use existing artifacts and implement the task list.test-only: run verification on the current feature branch and reportresults.
prd-only: create or revise only tasks/{slug}/prd.md.spec-driven and ralph-loop are out of scope for this skill-first v1 unless they are rebuilt as explicit skill instructions.
~30 seconds. Free. No account. Every finding cites a rule and a line of evidence.