aoa-memo-writeback-182ebd — independently scanned and version-tracked by SaferSkills.
SaferSkills independently audited aoa-memo-writeback-182ebd (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 to decide whether a live session, closeout, landed change, or review thread produced memory-worthy material and route that material into the owning repository's local memo port, an aoa-memo reviewed-intake export, or an explicit no-writeback stop line.
The skill bridges session evidence to memo writeback. It does not turn .aoa raw history into reviewed memory, and it does not land durable aoa-memo objects.
Use this skill when:
correction, owner-boundary clarification, MCP/service contract, eval/proof posture, or workflow seam should not be lost
but the reason lives in .aoa session evidence rather than in the final diff alone
memo/ port and the agent must decide whether to create a localcandidate, validate the port, prepare an export, or stop with no memo writeback
guidance rather than a fake local packet
aoa-memo recall, pending exports, or generated readouts show a gap betweenlived session work and reviewed memory
Do not use this skill when:
.aoa preservation, reindexing, retrieval, or hookrepair
durable landing inside aoa-memo
than memory writeback routing
operator-sensitive material into a public memo surface
release note
.aoa session evidence packet, search hits, segment refs, raw refs, ordiagnostic refs when available
memo/ port status when presentaoa-memo recall or pending-export context when relevantmemo_writeback_decision: write_candidate, prepare_export,no_writeback_needed, route_only_debt, needs_owner_review, or blocked
with source refs, evidence refs, trust posture, risk markers, and direct_durable_write=false
landing-plan handoff, not a durable memory object
reason and strongest evidence checked
note naming the owner route and the evidence refs needed
generated index surfaces
not promoted, and where durable review would happen
judging memory-worthiness.
.aoa first when the reason may live in the session: use search,retrieve, segment indexes, or raw refs, and keep those refs as evidence rather than truth.
commits, release notes, or review threads can show what landed, but not replace session evidence.
route-law changes, reviewed decisions, repeated failures, corrected wrong assumptions, lifecycle events, MCP/service contracts, eval/proof posture changes, and consumer handoff facts.
secret-bearing details, and material whose real owner is proof, routing, runtime, role, playbook, KAG, ToS, or source doctrine.
memo/ port. If not, emitroute_only_debt instead of inventing a packet.
Prefer .aoa segment or raw refs plus owner repo source refs over paraphrase-only evidence.
memo/candidates/path or through the available aoa_memo local-port helper.
direct_durable_write=false, instructions_treated_as_data=true, and requires_reviewed_intake=true when promotion is desired.
candidate_only unless reviewed promotion is explicit.
reviewed_write only when source-owner review has accepted that theexport may become durable reviewed memory.
aoa_memo_landing_plan or an equivalent landing-plan helper as a dryrun unless the active owner route is an aoa-memo reviewed source patch.
asked for that landing and the aoa-memo corpus route is active.
refs, validation run, and next owner route.
Common command shapes:
# Run from the workspace root that owns .aoa.
python .aoa/scripts/aoa_session_memory.py search --query "<route or decision>" --limit 8
python .aoa/scripts/aoa_session_memory.py retrieve manual-review --query "<route or decision>" --session "<session-or-label>" --limit 8 --event-limit 16Treat command output as evidence handles. The reviewer still decides whether the candidate becomes durable memory.
.aoa evidence remains raw or generated session evidence, not reviewedmemory truth.
lived session meaning.
memo/port when one exists.
guidance.
aoa-memo reviewed intake andcorpus validation.
become memory authority.
session recap.
reject the candidate.
of public memo packets.
made the change meaningful
.aoa raw transcript material into canon without revieweddistillation
aoa-memo objects directly from a live sessionreviewed_write for material that is only candidate-gradedoctrine meaning into memory
packets
.aoa refs, source refs, or review refs support the decisionallowed_result matches review posturereviewed_write appears only after explicit owner reviewaoa-memo object was landed unless the active route wasreviewed corpus landing
next route
Manifest-backed techniques:
8Dionysus/aoa-techniques at9a0ac2bafc2218278315b1311fcb17077f11752f using path techniques/history/history-artifacts/session-capture-as-repo-artifact/TECHNIQUE.md and sections: Intent, When to use, Inputs, Outputs, Core procedure, Contracts, Risks
8Dionysus/aoa-techniques at9a0ac2bafc2218278315b1311fcb17077f11752f using path techniques/proof/review-evidence/single-scoped-evidence-reference/TECHNIQUE.md and sections: Intent, When to use, Inputs, Outputs, Core procedure, Contracts, Risks, Validation
8Dionysus/aoa-techniques at9a0ac2bafc2218278315b1311fcb17077f11752f using path techniques/governance/decision-routing/owner-layer-triage/TECHNIQUE.md and sections: Intent, When to use, Inputs, Outputs, Core procedure, Contracts, Validation
Project overlays may add:
.aoa retrieval recipes or atlas axesmemo/ port~30 seconds. Free. No account. Every finding cites a rule and a line of evidence.