java-refactor-engram-contract — independently scanned and version-tracked by SaferSkills.
SaferSkills independently audited java-refactor-engram-contract (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 java-refactor-anchor-first and its workflow-private subagents when they exchange phase evidence through Engram.
Do not use it for generic Java refactoring advice, testing technique selection, or non-Engram workflows.
project and run_id before any Engram access; do not infer them from cwd or session.java-refactor-anchor-first/{run_id}/..., unless an exact caller allowlist explicitly overrides defaults.mem_search(query: topic_key, project: project, scope: "project") followed by mem_get_observation(id); search previews are never source material.mem_save, exact topic_key, scope: project, capture_prompt: false, and **What**/**Why**/**Where**/**Learned** content.state, coverage, mutation, review-strategy, evidence-report); use distinct topic keys for per-slice artifacts (tcr-slice-{n}).state; never overwrite it with a partial snapshot.| Condition | Action |
|---|---|
Missing project, run_id, or active-run validation | Block before Engram access. |
Topic key outside the run namespace or active_run_topic_prefix/allowed_topic_keys allowlist | Block as stale/mismatched context. |
| Required dependency topic missing or ambiguous | Block; do not proceed from preview text. |
Updating state | Read existing state first, then merge gate status, topic refs, attempts, retries, and human decisions. |
state (includes anchor_attempts[]), baseline-audit, target-scope, test-anchor, coverage, mutation, slice-plan, review-strategy, tcr-slice-{n}, evidence-report under java-refactor-anchor-first/{run_id}/.project: <required Engram project name>
run_id: <required stable run id>
phase: baseline | test-anchor | refactor | evidence
read_topics: [<exact dependency topic keys>]
write_topics: [<exact output topic keys>]
active_run_topic_prefix: <optional prefix for output-key validation>
allowed_topic_keys: [<optional exact allowlist for output-key validation>]
phase_context: <optional compact phase-specific payload>mem_search in project scope, then mem_get_observation for full content.status: blocked | ready | complete | failed
project: <provided Engram project name>
run_id: <stable run id>
engram_topics:
read: [<exact topic keys read>]
written: [<exact topic keys written>]
next_recommended: next_task | human_decision | caller_decides | none
human_question: <one question only, when blocked>
risk: low | medium | highPhase-specific fields such as gate, anchor_attempt, retry_decision, quality_gate, or final report details remain in each agent's own output contract.
| Phase | Agent | Reads | Writes | Owns |
|---|---|---|---|---|
| Baseline/tooling | baseline-auditor | state, target-scope (if provided) | baseline-audit, coverage, mutation | Build/test config audit, tooling readiness |
| Test anchor | test-anchorer | state, baseline-audit, target-scope, (coverage, mutation if available) | test-anchor, coverage, mutation | Characterization/unit-test anchors, target-scope coverage, mutation readiness |
| Refactor quality | tcr-worker | state, baseline-audit, target-scope, test-anchor, coverage, mutation, slice-plan, review-strategy | tcr-slice-{n} | One behavior-preserving refactor slice with quality gates |
| Evidence | evidence-curator | state, baseline-audit, target-scope, test-anchor, coverage, mutation, slice-plan, review-strategy, tcr-slice-{n} | evidence-report | Final gate matrix and traceable reporting |
| Orchestration | primary | Subagent envelopes + compact gate evidence (test-anchor, coverage, mutation) | state, target-scope, slice-plan, review-strategy | Gate advancement, retry decisions, human waivers |
~30 seconds. Free. No account. Every finding cites a rule and a line of evidence.