Skillforge Public Repo Sanitizer
Audit a repo for secrets, personal paths, client-specific references, and OSS-readiness gaps before publishing.
install
source · Clone the upstream repo
git clone https://github.com/jamiojala/skillforge
Claude Code · Install into ~/.claude/skills/
T=$(mktemp -d) && git clone --depth=1 https://github.com/jamiojala/skillforge "$T" && mkdir -p ~/.claude/skills && cp -r "$T/skills/public-repo-sanitizer" ~/.claude/skills/jamiojala-skillforge-public-repo-sanitizer && rm -rf "$T"
manifest:
skills/public-repo-sanitizer/SKILL.mdsource content
Public Repo Sanitizer
Audit a repo for secrets, personal paths, client-specific references, and OSS-readiness gaps before publishing.
Source: Advanced first-party pack
Use this skill when
- The request signals
or a directly related problem.open source - The request signals
or a directly related problem.public repo - The request signals
or a directly related problem.sanitize - The request signals
or a directly related problem.release - The request signals
or a directly related problem.publish
Gather this context first
- Relevant files, modules, or specs that define the current surface.
- Constraints, rollout limits, or non-goals that change the recommendation.
- What success looks like for the user, operator, or release owner.
Recommended workflow
- Confirm the trigger fit and boundaries before expanding scope.
- Identify the highest-risk files, interfaces, or decision points first.
- Produce a bounded plan or implementation slice with exact targets.
- Run the listed validation hooks or explain what blocks them.
- Return rollout, fallback, and open-question notes for the next agent or maintainer.
Output contract
- Capability summary and why this skill fits the request.
- Concrete file, module, or artifact targets.
- Validation plan and residual risk notes.
Failure modes to watch
- The pack matches the theme of the request but not the highest-leverage failure domain.
- Validation is mentioned without enough proof for another operator or agent to repeat it.
- The output becomes generic advice instead of a bounded next-step plan.
- Release hygiene checks exist but are too narrow to catch high-cost leaks.
- The pack flags noise instead of the operator-visible failure modes that matter.
Operational notes
- State the smallest safe slice that can be executed or reviewed next.
- Leave enough evidence behind that another maintainer can continue without re-deriving the workflow.
- Call out where human review or approval changes the recommended path.
- Tie hygiene checks to a concrete release gate or review step.
- Separate high-confidence blockers from advisory findings so the operator can act quickly.
Dependency and composition notes
- Let this pack lead only when it owns the main bottleneck; otherwise treat it as a specialist sidecar.
- If another pack has a narrower, more concrete surface, hand off with explicit files, risks, and validation goals.
- Pairs well with security, debugging, and docs packs to turn checks into operator-ready release gates.
Validation hooks
git_delegate_code_review
Model chain
deepseek-ai/deepseek-v3.2gemini-2.5-pro
Pack contents
for portable agent-skill usageSKILL.md
for runtime registry loadingskill.yaml
for human install and review contextREADME.md
for richer metadata and catalog indexingmarketplace.yaml