feat(registry): resolve + load registry assets into tasks (#246) - #4
Draft
Kalindi-Dev wants to merge 1 commit into
Draft
feat(registry): resolve + load registry assets into tasks (#246)#4Kalindi-Dev wants to merge 1 commit into
Kalindi-Dev wants to merge 1 commit into
Conversation
…s#246) Builds on the catalog PR to actually consume registry assets at task time: - Orchestrator resolve-step (`resolveRegistryAssets`): resolves a blueprint's `registry://` mcp_server / cedar_policy_module / skill refs at task start, fail-closed; stamps the `{kind,id,version}` triples on the TaskRecord for audit, merges resolved cedar_text into `cedar_policies`, and threads the runtime bundle into the agent payload. - Blueprint asset props + onUpdate fix: `assets.{mcpServers,cedarPolicyModules, skills}` with `RegistryRefValidation`; the three onUpdate helpers now write the asset-ref columns so redeploying an onboarded repo no longer drops them. - Agent loaders (registry.loader): mcp_server merges into `.mcp.json`; cedar_policy_module flows through PolicyEngine's unannotated `extra_policies`; skill prompt fragments append to the system prompt (build_skill_prompt_fragment). - TaskOrchestrator IAM: read-only bedrock-agentcore registry access so the orchestrator can resolve refs. Depends on the catalog PR (feat/246-registry-catalog): imports the RegistryClient port, ref grammar, and resolver from that branch.
Kalindi-Dev
pushed a commit
that referenced
this pull request
Aug 5, 2026
…amples#363) (aws-samples#364) * docs(design): add CI build performance design doc (aws-samples#363) Records how the build workflow's cost is distributed (//cdk:test was the ~91% long pole), why it was expensive (redundant type-check, fixed in aws-samples#357 / PR aws-samples#359), and the remaining levers (shard, coverage-gate, runner size, path-filter) with sequence and implementer notes — including the merge-queue required-check constraints that shape #2 and aws-samples#5. Companion to umbrella issue aws-samples#363; keeps the optimization roadmap as a durable, reviewable design artifact. Co-Authored-By: Claude Fable 5 <noreply@anthropic.com> * docs(design): add sharding design for CDK suite (aws-samples#363) Expands the CI build-performance doc with a full design for item #2 (shard the CDK suite): the jest --shard mechanism, a shards-vs-wall-time table showing fixed overhead dominates past ~4 shards, the fan-out + aggregate-gate pattern needed to keep `build` a single required context on merge_group, cross-shard coverage merge before threshold enforcement, and open questions (synth placement, deploy artifact, shard balance). Co-Authored-By: Claude Fable 5 <noreply@anthropic.com> * docs(design): correct CI build-performance figures + shard/priority analysis per review (aws-samples#363) Remediates @theagenticguy's CHANGES_REQUESTED (7 inline comments) with measured merge_group data from runs 30412723686 (Build 171s, //cdk:test 125.08s / 136 suites), 30409171662 (186s), 30400147506 (148s): 1. Ledger split so aws-samples#359 (type-check removal) and aws-samples#371 (Lambda-bundling disable) each get credit; restated current baseline (build ~148-186s, //cdk:test ~125s) with the run IDs it came from — not the stale ~346s/~298s post-#1 column. 2. 'Paid for twice' scoped: src/ went twice->once; test/ went once->zero (tsc only includes src/**); eslint tsconfig.dev.json backstops narrowly; noted deferred tsc --noEmit -p tsconfig.dev.json lever. 3. Idle-window claim corrected: synth ends +142s vs //cdk:test +171s -> ~29s idle, not ~200s; named //cdk:synth:quiet as the next binding constraint for sharding. 4. fixed_overhead restated as 120-220s measured from job start (not ~95s from install); added cache-miss finding (node_modules/venv/jest miss on every merge_group run) as new recommendation aws-samples#6. 5. Shard math recomputed with measured inputs -> 4-way lands at/above whole build; #2 demoted to deferred (revisit trigger //cdk:test > ~250s); sequencing re-ranked to elevate cache hit-rate (aws-samples#6), runner size (#4), path filters (aws-samples#5) over sharding. 6. Aggregate-job if: fixed to GitHub's documented pattern (job if: always(), gate moved into a step keying on failure||cancelled); explained why bare contains() never runs. 7. Check-run names corrected to build (agentcore) (build.yml compute_type matrix + ruleset 14980587, confirmed PR aws-samples#672); shard dim yields build (agentcore, N); flagged as a REQUIRED-CONTEXT change, not workflow-only. Rebased onto latest origin/main; regenerated Starlight mirror via docs:sync. Relates to aws-samples#363 Co-authored-by: Claude Opus 4.8 <noreply@anthropic.com> * docs(design): correct cache/disk-overhead analysis + add nav entries per re-review (aws-samples#363) Re-review round 2 (theagenticguy) surfaced 3 new inline findings plus a dropped nav finding. Corrections, cross-checked against per-step job timings: - Cache hits do not save time. Comparing a miss merge_group run (30412723686) vs a hitting pull_request run (30419878840) with the identical yarn.lock hash: Cache * restore 1s->9s, yarn 48.35s->55.09s, Install dependencies 56s->60s. The hit path is net SLOWER because mise.toml [tasks.install] runs yarn install --check-files, which re-verifies every restored file; uv sync installs 127 pkgs in 121ms so a venv cache cannot pay for itself. Rec aws-samples#6 reframed as a --check-files investigation and demoted in sequencing; no more cache-win claims. - Free Disk Space is the dominant addressable overhead, not cache misses. Per-step timings (gh api .../jobs/<id> .steps[]): run 30412723686 (job 90452344735) 148s/220s = 67%; run 30419878840 (job 90474296240) 97s/179s = 54%; run 30418865914 (job 90471207016) 75s/158s = 47%. Added as new ledger item aws-samples#7 ahead of aws-samples#6; feeds the shard math. - fixed_overhead breakdown named inline: of 120-220s, Free Disk Space is 75-148s and install is a cache-independent ~56-60s. ~250s shard trigger is now conditional on overhead staying 120-220s. - Nav fix: astro.config.mjs Architecture sidebar is an explicit list, so added slug entries architecture/ci-build-performance and architecture/bedrock-cost-attribution (both were missing from nav). Relates to aws-samples#363 Co-authored-by: Claude Opus 4.8 <noreply@anthropic.com> --------- Co-authored-by: bgagent <345885+scottschreckengaust@users.noreply.github.com> Co-authored-by: Claude Fable 5 <noreply@anthropic.com>
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Summary
Builds on the catalog PR to consume registry assets at task time.
resolveRegistryAssets): resolves a blueprint'sregistry://mcp_server/cedar_policy_module/skill refs at task start, fail-closed; stamps{kind,id,version}triples on the TaskRecord, merges resolved cedar_text intocedar_policies, threads the runtime bundle into the agent payload.assets.{mcpServers,cedarPolicyModules,skills}withRegistryRefValidation; the three onUpdate helpers now write the asset-ref columns so redeploying an onboarded repo no longer drops them.registry.loader): mcp_server →.mcp.json; cedar_policy_module → PolicyEngine unannotatedextra_policies; skill prompt fragments → system prompt.Test plan
mise run buildgreen (2595 tests total)--context forkBlueprintRepo=owner/repo; submit a task that pins all three asset kinds01KYJ5FT0RDZJ72J7R5BKCDM38)