Adk Memory Architecture

ADK memory architecture

Deliver a focused implementation, adaptation, audit or decision for the target agent's memory architecture. Keep the target's domain, interfaces, package manager and dependency pins. The book and companion repository are not required.

Inspect the actual boundary

1. Read project instructions, manifests and lockfiles, configuration templates, application startup, agent/Runner assembly, tools, storage adapters and tests. Identify the requested outcome; an audit or recommendation does not authorise implementing every discovered gap. Inspect imports and test setup before executing them: either can initialise remote clients or load credentials.
2. Use the target's existing interpreter. Optionally run the bounded, read-only inspector: "$PYTHON" "$SKILL_DIR/scripts/inspect_project.py" --project "$PROJECT_DIR". Set these variables to the selected interpreter, installed skill directory and target project. Read compatibility.md for its exit codes, coverage and version checks. It reports signals, not proof of correct authorisation. Inspect unresolved pins and unscanned files manually.
3. Map authenticated user/tenant/customer identity, session ownership, state scopes, exact profiles, memory submission and retrieval, document releases, structured data freshness, replay, telemetry and erasure. Record which controls are constructed by the real serving entrypoint. Importing a bare agent does not prove its API startup protections are active.
4. Record configuration key names and provenance without exposing values. Root and nested configuration, CLI account, ADC quota project, model API key and deployed workload identity can differ. Missing credentials block live verification; continue safe local implementation and offline checks.

Select the mode from the fact's meaning

Load only the relevant reference; combine modes when the requirement needs it.

Use the owning transactional service for current entitlements or business mutations. Exact user-controlled settings belong in a deterministic profile. Neither semantic memory nor retrieved prose authorises an action. Do not add all four stores merely because they are available.

When implementing the HTTP/ADK boundary, read integration-recipes.md for concrete assembly, trusted-context, event and replay contracts. When connecting or hosting managed backends, read managed-deployment.md. For failed setup, retrieval, protection, probes or cleanup, use failure-recovery.md before proposing a retry.

Implement or adapt

1. Present a short plan before broad changes: affected seams/files, identity and lifetime contracts, failure behaviour and observable acceptance tests. Preserve custom implementations that already meet those contracts. Make the smallest coherent change; do not replace the app with a chapter demo.
2. Derive scope in trusted code, outside model arguments and client state. Reauthorise each data access. Protect material before model exposure and before persistence, then project public output separately. A later output rejection cannot undo a tool's earlier side effect.
3. Follow the chosen reference's implementation sequence and error cases. Treat recommendations labelled production additions as work needing implementation and tests. Do not describe them as inherited guarantees. Preserve pending and indeterminate writes and their operation identities; never turn a timeout into an automatic fresh submission.
4. Keep models, resource names, locations, timeouts, retention and budgets configurable. Use ADC/workload identity and managed secrets where relevant. Do not place credentials in memory, events, fixtures or logs. Configuration examples contain synthetic placeholders, never copied project values.

Permission and lifecycle

Inspection is read-only by default. Before enabling APIs, creating resources or secrets, changing IAM, migrating data, deploying, deleting, increasing quotas or running paid/destructive tests, show the exact project, regions, resources, identities, commands and bounded work allowance, then obtain explicit approval. Code-generation permission is not deployment permission. Reuse an existing approval only for its exact scope; continue independent offline work while a required decision is pending. Never use Owner, Editor or wildcard grants as a troubleshooting shortcut.

Read operations-validation.md before cloud work, recovery or cleanup. Use isolated test targets and synthetic data, idempotent setup and an ownership journal. Cleanup is separate, requires confirmation, and may target only resources created by this workflow. Shared regional RAG infrastructure is outside routine corpus cleanup. Bound retries, polling and total elapsed work; do not bypass screening to recover availability.

Validate and report

1. Run the target's applicable offline tests and checks using its existing environment. Exercise the actual boundary through deterministic doubles: cross-user access, consent denial, duplicate submission/replay, expiry, unknown retrieval sources, stale structured data and protection failure as relevant. Assert stored effects as well as returned text. Include failure and retry paths; a health response is not a model or persistence test.
2. Inspect the second invocation: no duplicate files, configuration churn, repeated side effects or untracked operations. For production additions, test worker loss, lease expiry and late writes at the relevant seam.
3. Run live acceptance only within approved scope. Report each phase as passed, failed, blocked or not run. Separate local doubles, real managed services from a local API, and deployed hosted behaviour. Historical partial acceptance does not certify a target deployment.
4. Finish with files changed, commands actually executed and exit/results, exact versions, evidence type, unverified behaviour and precise manual or approval steps. For an audit, pair each gap with a concrete remedy and test. Use provenance.md when the origin of a rule matters; skill-validation.md records this package's own checks and limits.

Independent community project; not affiliated with or endorsed by Google.