refactorch-phases — independently scanned and version-tracked by SaferSkills.
SaferSkills independently audited refactorch-phases (Agent Skill) and scored it 100/100 (green). The audit ran 55 deterministic rules across Security, Supply Chain, Maintenance, Transparency, and Community; it found 0 high-severity and 0 lower-severity findings. The full rule-by-rule trace and per-finding evidence are below. Free, methodology-open.
Findings & checks · 0 flagged
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 for refactorch primary and workflow-private subagents when they exchange compact phase artifacts through Engram.
Do not use it for generic refactoring advice, code-smell analysis, Project Profile creation steps, or non-Engram workflows.
project, exact topic_key, scope: "project", type: "architecture", and capture_prompt: false.For each required artifact reference:
mem_search(query: "<exact topic_key>", project: "<project>", scope: "project")
mem_get_observation(id: <matched observation id>)Block before phase work if a required artifact is missing or ambiguous. Do not proceed from search preview text.
After retrieving the full observation, block if it does not correspond to the exact requested topic_key and project.
For each generated artifact:
mem_save(
title: "<topic_key>",
topic_key: "<topic_key>",
type: "architecture",
project: "<project>",
scope: "project",
capture_prompt: false,
content: "<full compact Markdown artifact>"
)topic_key enables upserts. Saving the same reusable or evolving artifact again updates the latest artifact instead of creating a separate handoff contract.
refactorch/project-profile/{repo-key}
refactorch/runs/{run-id}/target-brief
refactorch/runs/{run-id}/refactor-plan
refactorch/runs/{run-id}/gate-reviewrepo-key should be a short stable repository identifier, such as the repository folder name or remote slug.
target-brief, refactor-plan, and gate-review are run-scoped. project-profile is reusable and repository-scoped.
The refactorch primary owns target-brief. It writes this artifact after identifying the requested refactor target and before launching later planning phases.
Minimal Markdown shape:
# Target Brief: <target-name>
## Request
- User request:
- Target path or symbol:
## Scope
- In scope:
- Out of scope:
## Known Context
- Project Profile reference:
- Important caller/test hints, if already known:
## Open Questions
- <one question only, or None>Every RefactorCh phase receives the smallest useful handoff:
project: <required Engram project name>
repo_path: <absolute repository path>
repo_key: <stable repository identifier>
run_id: <optional for reusable project-profile, required for run-scoped artifacts>
read_artifacts:
- kind: project-profile | target-brief | refactor-plan | gate-review
topic_key: <exact Engram topic key>
write_artifact:
kind: project-profile | target-brief | refactor-plan | gate-review
topic_key: <exact Engram topic key>
phase_context: <optional compact phase-specific payload>Block before Engram access when project, repo_path, repo_key, or a required topic_key is missing.
Use artifact references instead of expanded documents when handing off between agents:
artifact_ref:
kind: project-profile | target-brief | refactor-plan | gate-review
project: <Engram project name>
topic_key: <exact Engram topic key>
title: <document title>If a document is short enough and useful for the next prompt, the caller may include it as context, but the durable handoff remains the Engram artifact reference.
Every phase returns this SDD-style envelope:
status: blocked | ready | complete | failed
project: <Engram project name>
executive_summary: <1-3 sentence human-readable summary>
artifacts:
- kind: project-profile | target-brief | refactor-plan | gate-review
project: <Engram project name>
topic_key: <Engram topic key>
title: <document title>
next_recommended: scout | planner | gatekeeper | human_decision | none
risks:
- <compact risk, or "None">
skill_resolution: injected | fallback-registry | fallback-path | noneskill_resolution follows the SDD convention: injected means project standards were provided by the orchestrator, fallback-registry means the agent loaded them from the registry, fallback-path means the agent loaded explicit skill paths, and none means no extra skills were loaded.
next_recommended is v1-only. Future executor or auditor phases must extend this enum deliberately instead of overloading existing values.
Add phase-specific fields only when they are essential for routing. Do not include expanded peer artifacts or raw evidence in the envelope.
refactorch/project-profile/{repo-key}.refactorch/runs/{run-id}/... topic keys.| Shared skill owns | Agent files and phase skills own |
|---|---|
| Topic keys | Responsibility |
| Engram read/write protocol | Permissions and forbidden actions |
| Common input shape | Related skills |
| Shared result envelope | Phase-specific behavior |
| Artifact reference format | Filled-in artifact content values |
| Target brief, refactor plan, and gate review handoff contracts | Project Profile creation workflow, templates, and refresh triggers |
~30 seconds. Free. No account. Every finding cites a rule and a line of evidence.