miclaw-app 0.1.0

package/.next/standalone/.next/server/app/skills.segments/skills/__PAGE__.segment.rsc

@@ -0,0 +1,1097 @@
+1:"$Sreact.fragment"
+2:I[22450,["/_next/static/chunks/0mmpyn9fv2fjf.js","/_next/static/chunks/0d3shmwh5_nmn.js","/_next/static/chunks/12p3iqtw2ohfq.js"],"ExpandableBody"]
+d:I[97367,["/_next/static/chunks/0mmpyn9fv2fjf.js","/_next/static/chunks/0d3shmwh5_nmn.js"],"OutletBoundary"]
+e:"$Sreact.suspense"
+3:Tc6c,# Implement Ticket Pipeline
+
+You are orchestrating a full implementation pipeline for ticket: **$ARGUMENTS**
+
+Follow these phases strictly and in order. Use the custom subagents defined in `~/.claude/agents/` to do the work -- do NOT do the implementation or review yourself.
+
+---
+
+## Phase 0: Setup
+
+1. Determine the repo root: `git rev-parse --show-toplevel`
+2. Read the repo's instruction file (check in order: AGENTS.md, CLAUDE.md, .clauderules at repo root). You will pass this file path to each subagent so they know the standards.
+3. Fetch the ticket details for **$ARGUMENTS** using Linear (or parse the ticket ID to understand context if provided inline).
+4. Analyze the ticket requirements and determine which packages, services, or modules need changes.
+
+---
+
+## Phase 1: Implement
+
+**Goal**: Write code and tests for the ticket.
+
+### If changes span a single package/service:
+- Spawn ONE `implement` subagent with the ticket requirements, repo instruction file path, and target area.
+
+### If changes span multiple packages/services:
+- Spawn MULTIPLE `implement` subagents IN PARALLEL (one per package/service), each in its own **worktree** (`isolation: worktree`).
+- Each agent gets: the ticket requirements, repo instruction file path, and its specific package/service scope.
+- After all parallel agents complete, integrate their changes. If worktrees were used, merge the branches they created.
+
+Wait for all implementation work to complete before proceeding.
+
+---
+
+## Phase 2: Review
+
+**Goal**: Code review of all changes.
+
+- Spawn the `review` subagent to review ALL changes from Phase 1.
+- The review agent will read the repo instruction file and `~/.claude/review-instructions.md` (if it exists) automatically.
+- Wait for the review to complete and read its findings.
+
+### Decision point:
+- If the review finds **no blocking issues**: skip Phase 3, go directly to Phase 4.
+- If the review finds **blocking issues**: proceed to Phase 3.
+
+---
+
+## Phase 3: Address Review (only if needed)
+
+**Goal**: Fix all blocking issues from the review.
+
+### If fixes are isolated to a single package/service:
+- Spawn ONE `address-review` subagent with the review feedback.
+
+### If fixes span multiple packages/services:
+- Spawn MULTIPLE `address-review` subagents IN PARALLEL, each scoped to its package and the relevant review comments.
+
+Wait for all fixes to complete before proceeding.
+
+---
+
+## Phase 4: Fix Tests
+
+**Goal**: Ensure all tests pass.
+
+- Spawn the `fix-tests` subagent to run the test suite and fix any failures.
+- If the repo has multiple test suites (e.g., per-package), spawn MULTIPLE `fix-tests` subagents IN PARALLEL.
+- Wait for all tests to pass.
+
+If tests cannot be fixed after the agent's attempts, report the failures to the user rather than looping.
+
+---
+
+## Phase 5: Summary
+
+After all phases complete, provide a concise summary:
+
+- **Ticket**: $ARGUMENTS
+- **What changed**: list of files/packages modified
+- **Tests**: pass/fail status
+- **Review**: whether all issues were addressed
+- **Ready for**: what the user should do next (e.g., "ready for git staging and commit")
+
+Do NOT run `git add` or `git commit`. The user handles all git operations.0:{"rsc":["$","$1","c",{"children":[["$","div",null,{"className":"mx-auto max-w-5xl px-8 py-10","children":[["$","div",null,{"className":"mb-8","children":[["$","div",null,{"className":"flex items-baseline gap-3","children":[["$","h1",null,{"className":"text-2xl font-medium tracking-tight","children":"Skills"}],["$","span",null,{"className":"text-sm text-text-muted","children":7}]]}],["$","p",null,{"className":"mt-1 text-sm text-text-muted","children":"Reusable prompts and workflows"}]]}],["$","div",null,{"className":"mb-8","children":[["$","h2",null,{"className":"text-xs font-medium text-text-dim uppercase tracking-wide mb-3","children":"Global"}],["$","div",null,{"className":"space-y-3","children":[["$","div","/Users/gabry/.claude/skills/implement-ticket/SKILL.md",{"className":"border border-border rounded-sm p-5 ","children":[["$","div",null,{"className":"flex items-start justify-between gap-3","children":["$","div",null,{"className":"flex items-center gap-2 flex-wrap","children":[["$","h3",null,{"className":"text-sm font-medium","children":"implement-ticket"}],["$","span",null,{"className":"inline-flex items-center px-2 py-0.5 text-xs font-medium rounded-sm bg-surface-raised text-text-dim","children":"Global"}],null,"$undefined"]}]}],["$","p",null,{"className":"mt-1.5 text-sm text-text-muted leading-relaxed","children":"Full implementation pipeline for a ticket -- implement, review, address feedback, fix tests. Orchestrates subagents with parallelism for multi-package changes."}],"$undefined",["$","div",null,{"className":"mt-3 flex flex-wrap gap-2","children":[["$","span",null,{"className":"inline-flex items-center px-2 py-0.5 text-xs font-medium rounded-sm bg-accent/15 text-accent","children":"user-invocable"}],["$","span",null,{"className":"inline-flex items-center px-2 py-0.5 text-xs font-medium rounded-sm bg-surface-raised text-text-dim","children":"manual only"}],["$","span",null,{"className":"inline-flex items-center px-2 py-0.5 text-xs font-medium rounded-sm bg-surface-raised text-text-muted","children":"<ticket-id>"}],false,"$undefined"]}],["$","div",null,{"className":"mt-2 flex flex-wrap gap-x-4 gap-y-1 text-xs text-text-dim","children":["$undefined","$undefined","$undefined","$undefined"]}],["$","div",null,{"className":"mt-3","children":[["$","p",null,{"className":"text-xs text-text-dim mb-1.5","children":"Orchestrates:"}],["$","div",null,{"className":"flex flex-wrap gap-1.5","children":[["$","span","address-review",{"className":"inline-flex items-center px-2 py-0.5 text-xs font-medium rounded-sm bg-accent/15 text-accent","children":"address-review"}],["$","span","fix-tests",{"className":"inline-flex items-center px-2 py-0.5 text-xs font-medium rounded-sm bg-accent/15 text-accent","children":"fix-tests"}],["$","span","implement",{"className":"inline-flex items-center px-2 py-0.5 text-xs font-medium rounded-sm bg-accent/15 text-accent","children":"implement"}],["$","span","review",{"className":"inline-flex items-center px-2 py-0.5 text-xs font-medium rounded-sm bg-accent/15 text-accent","children":"review"}]]}]]}],["$","$L2",null,{"content":"$3"}]]}]]}]]}],"$L4"]}],["$L5"],"$L6"]}],"isPartial":false,"staleTime":300,"varyParams":null,"buildId":"BgLP6ChyOyE0k8Ahr8F2H"}
+7:Td51,# find-variant
+
+Looks up a genomic variant in the user's VCF file and researches it across public genomic databases.
+
+## Input Formats
+
+The user will provide a variant identifier in one of these formats:
+
+1. **chr:pos:ref:alt** -- e.g., `chr6:26092913:G:A`
+2. **rsID** -- e.g., `rs1800562`
+3. **ClinVar VCV ID** -- e.g., `VCV000000009` or a ClinVar URL
+
+## Instructions
+
+### Step 1: Resolve Coordinates
+
+Depending on the input format, resolve to GRCh38 genomic coordinates:
+
+- **chr:pos:ref:alt**: Use directly.
+- **rsID**: Fetch the dbSNP page `https://www.ncbi.nlm.nih.gov/snp/<rsID>` to get GRCh38 chr and position, plus ref/alt alleles.
+- **VCV ID**: Fetch the ClinVar variation page `https://www.ncbi.nlm.nih.gov/clinvar/variation/<ID>/` to get GRCh38 coordinates, gene, ref/alt, and any rsID.
+
+### Step 2: Check the VCF
+
+Find the VCF file. Look for `*.vcf.gz` files in the current working directory. If multiple exist, ask the user which one.
+
+Query using bcftools:
+
+```bash
+bcftools view -H <vcf_file> <chr>:<pos>-<pos>
+```
+
+Report whether the variant is present. If found, display the full VCF line and extract key fields:
+- Genotype (GT) -- translate to human-readable (e.g., "heterozygous", "homozygous alt", "homozygous ref")
+- Genotype Quality (GQ)
+- Read Depth (DP)
+- Allele Depths (AD)
+- Filter status (PASS or not)
+
+If NOT found, explicitly state the variant is absent from the VCF.
+
+### Step 3: Database Research
+
+Launch a research subagent to query public databases **in parallel** with the VCF lookup when possible. The subagent should search the web for:
+
+1. **dbSNP**: Allele frequencies across populations (gnomAD, ALFA, TOPMED). A variant with >5% global frequency is common and unlikely to be pathogenic for a rare Mendelian condition.
+
+2. **ClinVar**: Clinical significance classification, review status (star rating), associated conditions, number of submitters, and assertion criteria. Flag 0-star submissions with no assertion criteria as low-confidence. Flag any "pathogenic" classification on a variant with >5% population frequency as likely erroneous.
+
+3. **ClinGen**: Check if the gene has ClinGen gene-disease validity curations. Note if it is on the ACMG secondary findings list.
+
+4. **GWAS Catalog**: Any significant GWAS associations (trait, p-value, odds ratio).
+
+### Step 4: Report
+
+Present a concise summary with:
+
+1. **VCF status**: Present or absent, with key genotype fields if present
+2. **Full VCF line**: In a code block if present
+3. **Gene and consequence**: Gene name, variant type (missense, intronic, synonymous, etc.)
+4. **ClinVar**: Classification, stars, conditions -- with quality caveats where warranted
+5. **Population frequency**: Global and major population frequencies
+6. **ClinGen**: Gene-disease validity status, ACMG secondary findings status
+7. **GWAS**: Notable associations if any
+8. **Bottom line**: One-sentence clinical interpretation appropriate for the genotype observed
+
+## Important Notes
+
+- Always use `bcftools` for VCF queries, never grep/zgrep.
+- VCF files may not contain rsIDs -- always query by genomic position.
+- Be skeptical of 0-star ClinVar submissions, especially for common variants.
+- Distinguish between Mendelian pathogenicity (rare, high-penetrance) and GWAS risk alleles (common, low effect size).
+- If the variant is in a well-known gene (BRCA1, HFE, CFTR, etc.), note relevant compound heterozygosity considerations.4:["$","div",null,{"children":[["$","h2",null,{"className":"text-xs font-medium text-text-dim uppercase tracking-wide mb-3","children":"Project"}],["$","div",null,{"className":"space-y-3","children":[["$","div","/Users/gabry/Desktop/.claude/skills/find-variant/SKILL.md",{"className":"border border-border rounded-sm p-5 ","children":[["$","div",null,{"className":"flex items-start justify-between gap-3","children":["$","div",null,{"className":"flex items-center gap-2 flex-wrap","children":[["$","h3",null,{"className":"text-sm font-medium","children":"find-variant"}],["$","span",null,{"className":"inline-flex items-center px-2 py-0.5 text-xs font-medium rounded-sm bg-surface-raised text-text-muted","children":"Desktop"}],null,"$undefined"]}]}],["$","p",null,{"className":"mt-1.5 text-sm text-text-muted leading-relaxed","children":"Look up a genomic variant in a local VCF and public databases (ClinVar, dbSNP, ClinGen, GWAS)"}],"$undefined",["$","div",null,{"className":"mt-3 flex flex-wrap gap-2","children":[["$","span",null,{"className":"inline-flex items-center px-2 py-0.5 text-xs font-medium rounded-sm bg-accent/15 text-accent","children":"user-invocable"}],"$undefined","$undefined",false,"$undefined"]}],["$","div",null,{"className":"mt-2 flex flex-wrap gap-x-4 gap-y-1 text-xs text-text-dim","children":["$undefined","$undefined","$undefined","$undefined"]}],["$","div",null,{"className":"mt-3","children":[["$","p",null,{"className":"text-xs text-text-dim mb-1.5","children":"Orchestrates:"}],["$","div",null,{"className":"flex flex-wrap gap-1.5","children":[["$","span","review",{"className":"inline-flex items-center px-2 py-0.5 text-xs font-medium rounded-sm bg-accent/15 text-accent","children":"review"}]]}]]}],["$","$L2",null,{"content":"$7"}]]}],"$L8","$L9","$La","$Lb","$Lc"]}]]}]
+5:["$","script","script-0",{"src":"/_next/static/chunks/12p3iqtw2ohfq.js","async":true}]
+6:["$","$Ld",null,{"children":["$","$e",null,{"name":"Next.MetadataOutlet","children":"$@f"}]}]
+10:T130d,# fix-tests
+
+Analyzes and fixes test files to conform to platform-services testing standards defined in AGENTS.md.
+
+## Instructions
+
+You are tasked with reviewing and fixing test files to follow the platform-services testing standards defined in AGENTS.md.
+
+### Step 1: Identify Test Files
+
+Ask the user which test file(s) to fix, or analyze all test files in the current package if not specified.
+
+### Step 2: Review Against Standards
+
+Check each test file for:
+
+**1. Parametrization**
+- ❌ Multiple similar test methods testing same function with different inputs
+- ✅ Use `@pytest.mark.parametrize` to combine them
+
+**2. AWS Service Mocking**
+- ❌ Direct `mock.patch` on boto3 clients
+- ✅ Mock the `get_{service}` context manager from package's `aws.py`
+- ✅ Use `botocore.stub.Stubber` for AWS service responses
+- Exception: `dynamodb_helper` uses `moto`
+
+**3. Logger Mocking**
+- ❌ `mock_logger = AsyncMock()` or `MagicMock()`
+- ✅ Use real `BioscopeLogger` from conftest.py fixture
+
+**4. Fixture Usage**
+- ❌ Mocks defined inline in test methods
+- ✅ All mocks defined as pytest fixtures in conftest.py or test file
+- ✅ Use `pytest_mock.MockerFixture` for creating mocks
+- ❌ Same `mocker.patch("path.to.module")` repeated 3+ times across tests
+- ✅ Extract repeated patches to named fixtures in conftest.py
+
+**5. Async vs Sync Assertions**
+- ❌ `mock_async_func.assert_called_once_with()`
+- ✅ `mock_async_func.assert_awaited_once_with()` for AsyncMock
+- ✅ `mock_sync_func.assert_called_once_with()` for MagicMock
+
+**6. Mock Types**
+- ❌ Using wrong mock type for function
+- ✅ `AsyncMock` for async functions
+- ✅ `MagicMock` for sync functions
+
+**7. Assertions**
+- ❌ Individual field assertions (`assert result.id == "123"`, `assert result.name == "foo"`)
+- ✅ Compare entire objects: `assert result == ExpectedModel(id="123", name="foo")`
+- ✅ Include negative assertions (what should NOT happen)
+
+**8. Mock Depth**
+- ❌ Mocking internal functions multiple levels deep
+- ✅ Mock at most one level deep (mock direct dependencies)
+
+**9. Import Organization**
+- ❌ Relative imports in test files: `from .conftest import fixture`
+- ✅ Test directories should NOT have `__init__.py`
+- ✅ Use absolute imports from the package being tested
+
+### Step 3: Suggest Fixes
+
+For each issue found:
+1. Show the current code
+2. Explain the problem
+3. Provide the fixed version
+4. Apply the fix if user approves
+
+### Step 4: Verify
+
+After fixes:
+1. Run tests to ensure they still pass
+2. Check coverage is maintained at 100%
+3. Run lint and type-check
+
+## Key Patterns
+
+### Fixture Example
+```python
+# conftest.py
+@pytest.fixture
+def logger() -> BioscopeLogger:
+    return get_logger("package-name", uuid4())
+
+@pytest.fixture
+def mock_some_function(mocker: MockerFixture) -> AsyncMock:
+    return mocker.patch("package.module.some_function", new_callable=AsyncMock)
+```
+
+### Parametrized Test Example
+```python
+@pytest.mark.parametrize(
+    ("input_value", "expected"),
+    [
+        ("foo", True),
+        ("bar", False),
+        ("", False),
+    ],
+)
+def test_validation(input_value: str, expected: bool) -> None:
+    result = validate(input_value)
+    assert result == expected
+```
+
+### Extracting Repeated Patches Example
+```python
+# BAD - same patch repeated in multiple tests
+class TestSomething:
+    def test_one(self, mocker: MockerFixture):
+        mock_fetch = mocker.patch("mypackage.module.fetch_data")
+        mock_fetch.return_value = {"key": "value1"}
+        ...
+
+    def test_two(self, mocker: MockerFixture):
+        mock_fetch = mocker.patch("mypackage.module.fetch_data")
+        mock_fetch.return_value = {"key": "value2"}
+        ...
+
+# GOOD - extract to fixture
+@pytest.fixture
+def mock_fetch_data(mocker: MockerFixture) -> MagicMock:
+    return mocker.patch("mypackage.module.fetch_data")
+
+class TestSomething:
+    def test_one(self, mock_fetch_data: MagicMock):
+        mock_fetch_data.return_value = {"key": "value1"}
+        ...
+
+    def test_two(self, mock_fetch_data: MagicMock):
+        mock_fetch_data.return_value = {"key": "value2"}
+        ...
+```
+
+### AWS Mocking with Stubber Example
+```python
+@pytest.fixture
+async def s3_stub(mock_get_s3: MockType) -> AsyncGenerator[Stubber]:
+    session = Session()
+    async with session.client("s3", region_name="us-east-2") as s3_client:
+        stubber = Stubber(s3_client)
+        stubber.activate()
+
+        @asynccontextmanager
+        async def _fake_get_s3():
+            yield s3_client
+
+        mock_get_s3.side_effect = _fake_get_s3
+        yield stubber
+        stubber.assert_no_pending_responses()
+        stubber.deactivate()
+```
+
+## Execution
+
+When this skill is invoked:
+1. Read the test file(s)
+2. Analyze against all standards
+3. Present findings organized by issue type
+4. Apply fixes one by one with user approval
+5. Verify tests still pass with 100% coverage8:["$","div","/Users/gabry/Desktop/.claude/skills/fix-tests/SKILL.md",{"className":"border border-border rounded-sm p-5 ","children":[["$","div",null,{"className":"flex items-start justify-between gap-3","children":["$","div",null,{"className":"flex items-center gap-2 flex-wrap","children":[["$","h3",null,{"className":"text-sm font-medium","children":"fix-tests"}],["$","span",null,{"className":"inline-flex items-center px-2 py-0.5 text-xs font-medium rounded-sm bg-surface-raised text-text-muted","children":"Desktop"}],null,"$undefined"]}]}],["$","p",null,{"className":"mt-1.5 text-sm text-text-muted leading-relaxed","children":"Analyze and fix test files to follow platform-services testing standards"}],"$undefined",["$","div",null,{"className":"mt-3 flex flex-wrap gap-2","children":[["$","span",null,{"className":"inline-flex items-center px-2 py-0.5 text-xs font-medium rounded-sm bg-accent/15 text-accent","children":"user-invocable"}],"$undefined","$undefined",false,"$undefined"]}],["$","div",null,{"className":"mt-2 flex flex-wrap gap-x-4 gap-y-1 text-xs text-text-dim","children":["$undefined","$undefined","$undefined","$undefined"]}],["$","div",null,{"className":"mt-3","children":[["$","p",null,{"className":"text-xs text-text-dim mb-1.5","children":"Orchestrates:"}],["$","div",null,{"className":"flex flex-wrap gap-1.5","children":[["$","span","fix-tests",{"className":"inline-flex items-center px-2 py-0.5 text-xs font-medium rounded-sm bg-accent/15 text-accent","children":"fix-tests"}],["$","span","review",{"className":"inline-flex items-center px-2 py-0.5 text-xs font-medium rounded-sm bg-accent/15 text-accent","children":"review"}]]}]]}],["$","$L2",null,{"content":"$10"}]]}]
+11:T2126,Create a weekly status report that contains the last 7 days of activity.
+
+There will be Four components - GitHub behavior, Notion page scraping, ASKING for any additional major items, and then styling the report. Each has a sub section below.
+
+The final output should be direclty to the CLI. Not to slack, or to Notion, or anywhere else.
+
+
+## Nick Gabry - GitHub Activity 
+
+Use this guide to fetch a summary of a GitHub user's activity over the last N days using the `gh` CLI.
+
+## Prerequisites
+
+- `gh` CLI installed and authenticated
+- Target GitHub username and date range
+
+## Commands
+
+Replace `<USERNAME>` and `<SINCE_DATE>` (format: `YYYY-MM-DD`) in each command.
+
+### 1. PRs Authored
+
+```bash
+gh search prs --author <USERNAME> --created ">= <SINCE_DATE>" \
+  --json number,title,state,repository,createdAt,url --limit 50
+```
+
+### 2. PRs Reviewed
+
+```bash
+gh search prs --reviewed-by <USERNAME> --updated ">= <SINCE_DATE>" \
+  --json number,title,state,repository,updatedAt,url --limit 50
+```
+
+### 3. Issues Authored
+
+```bash
+gh search issues --author <USERNAME> --created ">= <SINCE_DATE>" \
+  --json number,title,state,repository,createdAt,url --limit 50
+```
+
+### 4. Issues Commented On
+
+```bash
+gh search issues --commenter <USERNAME> --updated ">= <SINCE_DATE>" \
+  --json number,title,state,repository,updatedAt,url --limit 50
+```
+
+## LLM Prompt Template
+
+To get a formatted summary, pass the raw JSON output from all four commands to an LLM with the following prompt:
+
+```
+Given the following GitHub activity data for user <USERNAME> from <SINCE_DATE> to <TODAY>:
+
+PRs Authored:
+<paste JSON>
+
+PRs Reviewed:
+<paste JSON>
+
+Issues Authored:
+<paste JSON>
+
+Issues Commented On:
+<paste JSON>
+
+Summarize the activity in markdown. Include:
+- A table of PRs authored grouped by repository, with status (open/merged/closed)
+- A grouped summary of PRs reviewed, highlighting themes or feature areas
+- Any issue activity
+- A short TL;DR paragraph at the end
+```
+
+## Notes
+
+- The `--reviewed-by` filter returns PRs updated in the window, not necessarily reviewed in it. Some older PRs may appear if they were recently re-touched (e.g. bulk merges).
+- Use `--limit 100` if the user is highly active.
+- The `gh search` commands require GitHub's search index, which may lag by a few minutes for very recent activity.
+- To scope to a specific org or repo, add `--owner <ORG>` or `--repo <ORG>/<REPO>`.
+
+
+## Notion Page Scraping
+
+Ask the user to copy and paste their work from the week from Notion.
+
+
+## Status Report Style Guide                                                                                                                                         
+                                                         
+  ## Overview                                                                                                                                                                      
+                                                                                                                                                                                   
+  Status reports are written for a technical audience that is already familiar with the project                                                                                    
+  domain. The goal is to communicate what happened, what is in progress, and what is blocked —                                                                                     
+  clearly and without padding. Write like an engineer, not a PR person.                                                                                                            
+
+  ---
+
+  ## Structure
+
+  Reports are organized into **domain-grouped sections**, each with a bolded header and
+  indented sub-bullets. Common section headers:
+
+  - **WGS**
+  - **PGX**
+  - **PRS** (or **DataTecnica / PRS**)
+  - **Microbiome** (or **cmbio / microbiome**)
+  - **Hiring**
+  - **Misc**
+
+  Use flat bullets (no grouping) only when the week's work does not naturally cluster into
+  distinct tracks. As project surface area grows, prefer grouped sections.
+
+  ---
+
+  ## Formatting Rules
+
+  - Top-level items use `•` bullets
+  - Sub-items use `◦` indented under the parent bullet
+  - Section headers are bolded inline on the parent bullet line:
+    `• **WGS**` followed by `◦` sub-bullets
+  - Do not use headers (`#`, `##`) — everything lives in bullet form
+  - Links are inline and labeled: `[Slide Deck](url)` or `[architecture diagram](url)`
+  - No horizontal rules, no tables, no numbered lists
+  - No sign-off, no greeting, no date header
+
+  ---
+
+  ## Tone and Voice
+
+  - Terse and direct — omit filler words and transition phrases
+  - Drop the subject ("Built out core of...", not "I built out core of...")
+  - No editorializing or self-congratulation, with rare exceptions for genuinely notable wins
+    (e.g., "Big win to have this stood up" — used sparingly)
+  - Acknowledge blockers and friction plainly: "Communication here has gone unnervingly quiet"
+  - Name collaborators and external parties specifically — avoid vague references like "the team"
+  - Refer to external vendors, partners, and contacts by first name or company name consistently
+
+  ---
+
+  ## Content Norms
+
+  **What to include:**
+  - Work completed, with enough specificity that a technical reader understands the scope
+  - Work in progress, with a clear sense of where it stands and what is left
+  - Blockers or dependencies on external parties
+  - Notable decisions or pivots, briefly explained
+  - Hiring activity (interviews, pipeline stage, outcomes)
+  - Cross-functional collaboration that produced a concrete output
+
+  **What to omit:**
+  - Process narration ("I spent time thinking about...")
+  - Praise of others' work (that belongs in other channels)
+  - Business-context backstory the reader already knows
+  - Vague status like "making progress" without a concrete anchor
+  - Meeting attendance without an associated outcome or decision
+
+  ---
+
+  ## Technical Language
+
+  Use domain-specific terminology without explanation:
+  - Bioinformatics: VCF, WGS, SNP, multi-allelic, zygosity, heteroplasmy, haplotype,
+    PRS, PGX, CPIC, ClinVar, ClinGen, HGNC, dbSNP, pharmcat, rsID
+  - Infrastructure: lambda, Glue job, S3, Step Function, DynamoDB, Terraform, EFS
+  - Engineering: RFC, ingestion pipeline, docker, LLM, RAG, agentic, annotation
+
+  Acronyms do not need expansion. If a specific tool, catalog, or standard is referenced,
+  name it directly (e.g., "PGScatalog", "ClinPgx", "MONDO ontology").
+
+  ---
+
+  ## Status Signaling
+
+  Use plain language to convey completion state inline — do not use status tags or labels:
+
+  | State | Example phrasing |
+  |---|---|
+  | Done | "Checked in", "Finalized", "Live in prod", "Wrapped up" |
+  | Nearly done | "Entering validation and polishing phase", "Basically finished" |
+  | In progress | "Working on", "Started", "WIP", "Continuing" |
+  | Blocked / waiting | "Still waiting on handoff from DT team", "Communication has gone quiet" |
+  | Deprioritized | "Light [topic] work", "On back burner" |
+
+  ---
+
+  ## Length
+
+  A typical report is **4–8 top-level bullets**. Grouped sections with sub-bullets can
+  expand this, but total length should still be scannable in under two minutes. If a section
+  had no meaningful activity, omit it rather than noting "no updates" — unless the silence
+  itself is worth flagging (e.g., a blocked external dependency).
+
+  ---
+
+  ## Example Output should be in MARKDOWN format, using '*' as bullets
+
+* WGS
+    * Data Histories & Lineage system is now live in dev and prod, giving us an easy path
+      forward to setting up an agentic QA / validation flow and improved tracing of data
+      provenance for our production audit trail.
+    * Plan & ticket out the work for the QA / validation flow.
+    * Discussions and planning around short-term and long-term PRS improvements. Made a
+      light UI change on the PRS tile for the short-term.
+* Microbiome
+    * Continue to push cmbio on integration timeline and other issues. Got our S3 connection
+      set up — which doesn't sound like all that big a deal, but it was a bit like pulling
+      teeth to get this far.
+    * Assist the work Marie is doing whenever possible.
+* Misc
+    * Several MAB member genomics reviews with Chris. Reception continues to be very positive
+      and the real-time feedback continues to be extremely helpful.
+    * Set up the first wave of interviews next week for bioinfo roles.9:["$","div","/Users/gabry/Desktop/.claude/skills/make-status-report/SKILL.md",{"className":"border border-border rounded-sm p-5 ","children":[["$","div",null,{"className":"flex items-start justify-between gap-3","children":["$","div",null,{"className":"flex items-center gap-2 flex-wrap","children":[["$","h3",null,{"className":"text-sm font-medium","children":"make-status-report"}],["$","span",null,{"className":"inline-flex items-center px-2 py-0.5 text-xs font-medium rounded-sm bg-surface-raised text-text-muted","children":"Desktop"}],null,"$undefined"]}]}],["$","p",null,{"className":"mt-1.5 text-sm text-text-muted leading-relaxed"}],"$undefined",["$","div",null,{"className":"mt-3 flex flex-wrap gap-2","children":[["$","span",null,{"className":"inline-flex items-center px-2 py-0.5 text-xs font-medium rounded-sm bg-accent/15 text-accent","children":"user-invocable"}],"$undefined","$undefined",false,"$undefined"]}],["$","div",null,{"className":"mt-2 flex flex-wrap gap-x-4 gap-y-1 text-xs text-text-dim","children":["$undefined","$undefined","$undefined","$undefined"]}],false,["$","$L2",null,{"content":"$11"}]]}]
+12:T3684,# Create Validation Notebook Skill
+
+You are a WGS validation workflow assistant. Your job is to interactively create a Jupyter notebook for validating WGS pipeline changes.
+
+## Your Workflow
+
+When this skill is invoked, guide the user through creating a validation notebook by gathering the following information:
+
+**IMPORTANT**:
+- Use a SINGLE `AskUserQuestion` call to ask all 4 questions at once. Do NOT ask questions one at a time.
+- After gathering requirements, proceed DIRECTLY to generating and saving the notebook WITHOUT asking for user permission to proceed. The user invoked this skill to create a notebook, so just do it.
+
+### 1. Gather Requirements
+
+Use the `AskUserQuestion` tool to gather ALL inputs in a SINGLE call. The tool supports asking 1-4 questions at once - use this to collect all requirements together, not one at a time.
+
+**Question 1: Which packages were modified?**
+- Use `AskUserQuestion` with `multiSelect: true` to allow multiple selections
+- Provide the 4 most common packages as options (tool limit is 2-4 options):
+  - `genomics-wgs-query-service` - "WGS Query Service (Lambda: processWgsVariants, computeGeneticFeatures)"
+  - `genomics-variant-assessment-service` - "Variant Assessment Service (Step Function)"
+  - `genomics-prs-io-service` - "PRS IO Service (Step Function)"
+  - `genomics-prs-processor` - "PRS Processor (Step Function)"
+- The tool automatically provides an "Other" option where users can type additional packages
+- In "Other", users can specify: `vcf-processor`, `patient-vcf-to-iceberg`, or any other package (comma-separated)
+- If user provides upstream packages (vcf-processor, patient-vcf-to-iceberg) via "Other", inform them these require triggering full ingestion via wgs-ingestion-dev-utils
+- Parse any comma-separated packages from the "Other" field
+
+**Question 2: Which validation patients?**
+- Use `AskUserQuestion` with 2 options:
+  - Option 1 (Recommended): "All 20 validation samples" - Use all samples from VALIDATION_SAMPLES
+  - Option 2: "Specific patient IDs" - User will provide specific UUIDs
+- If user selects "Other" to provide specific patient IDs, parse the comma-separated UUIDs from their response
+
+**Question 3: PR URL (optional)**
+- Use `AskUserQuestion` with a single option:
+  - "No PR URL" - Skip this field (will use N/A)
+- The tool automatically provides "Type something" option for users to enter PR URL
+- If user types a URL in "Other", use it; otherwise use "N/A"
+
+**Question 4: Ticket ID (optional)**
+- Use `AskUserQuestion` with a single option:
+  - "No ticket ID" - Skip this field (will use N/A)
+- The tool automatically provides "Type something" option for users to enter ticket ID
+- If user types a ticket ID in "Other", use it; otherwise use "N/A"
+
+### 2. Generate Notebook from Template
+
+**CRITICAL REQUIREMENT**: For each modified package, you MUST create a SEPARATE code cell with a single `execute_package()` call. Do NOT combine multiple packages into one cell. Users need to run packages sequentially and wait for Step Function completion between invocations.
+
+Use the template notebook at `.claude/skills/create-validation-notebook/template.ipynb`:
+
+1. **Read the template**:
+   ```python
+   import json
+   with open(".claude/skills/create-validation-notebook/template.ipynb") as f:
+       notebook = json.load(f)
+   ```
+
+2. **Customize the metadata cell** (first cell):
+   - Replace `{TIMESTAMP}` with current timestamp
+   - Replace `{TICKET_ID}` with ticket ID or "N/A"
+   - Replace `{PR_URL}` with PR URL or "N/A"
+   - Replace `{PACKAGE_LIST}` with markdown list of modified packages
+   - Replace `{PATIENT_INFO}` with list of patient IDs and sample IDs only
+
+3. **Customize configuration cells**:
+   - Replace `{PATIENT_IDS}` with actual patient UUID list
+     - **CRITICAL**: Wrap each UUID string with `UUID()` constructor
+     - Example: `[UUID("990051c4-257e-4ee9-95d1-067f0fd6e4d9"), UUID("91e0ca25-d81d-4cf1-ae73-cfde9fd1f56b")]`
+     - NOT just strings: `["990051c4-257e-4ee9-95d1-067f0fd6e4d9", "91e0ca25-d81d-4cf1-ae73-cfde9fd1f56b"]`
+   - Replace `{PACKAGES_MODIFIED}` with list of modified package names (these stay as strings)
+
+4. **Generate package execution cells** (CRITICAL - one cell per package):
+   - For EACH modified package (user provided list), create SEPARATE cells:
+     - Markdown cell with package name and warning about sequential execution
+     - Code cell with `execute_package()` call for that ONE specific package
+   - **IMPORTANT**: Do NOT combine multiple packages into a single cell
+   - **IMPORTANT**: Each package MUST have its own dedicated code cell
+   - This ensures users run packages one at a time and wait for completion between invocations
+   - Remove the template package cells (marked with `{PACKAGE_NAME}`)
+   - Note: The notebook will analyze affected packages at runtime
+
+5. **Keep the diff cells as-is from the template** (cells 13 onward):
+   - The template already contains the correct per-product-group diff structure — do NOT modify these cells
+   - The diff section has separate cells for: download, variants, genomics/tiles, PRS, VCF, and LLM review
+   - Each cell calls `run_all_diffs(..., product_types={...})` with the relevant `DataProductType` set
+   - Users can run only the product groups relevant to their change, skipping others
+
+6. **Save the customized notebook**:
+   - Filename format: `YYYY-MM-DD_HH-MM-SS_validation.ipynb`
+   - Example: `2026-02-05_17-10-38_validation.ipynb`
+   - Get timestamp using bash: `date +"%Y-%m-%d_%H-%M-%S"`
+   - **CRITICAL**: Use this exact format - ISO date (YYYY-MM-DD), underscore, time (HH-MM-SS), underscore, "validation.ipynb"
+   - Save to: `packages/wgs_validations/` directory
+   - **IMPORTANT**: Do NOT ask for user permission before writing the file - just save it directly
+   - The user invoked this skill specifically to create a notebook, so proceed without confirmation
+
+### Template Placeholders
+
+Replace these placeholders in the template:
+
+- `{TIMESTAMP}` - Current date/time
+- `{TICKET_ID}` - Linear ticket ID or "N/A"
+- `{PR_URL}` - GitHub PR URL or "N/A"
+- `{PACKAGE_LIST}` - Markdown bullet list of modified packages
+- `{AFFECTED_PACKAGES_INFO}` - Explanation of affected packages and dependencies
+- `{PATIENT_INFO}` - List of patient IDs and sample IDs only (no population)
+- `{PATIENT_IDS}` - Python list of UUID objects (e.g., `[UUID("..."), UUID("...")]`)
+- `{PACKAGES_MODIFIED}` - Python list of package name strings (e.g., `["genomics-wgs-query-service"]`)
+- `{PACKAGE_NAME}` - Specific package name (duplicate cells for each package)
+- `{DEPENDENCIES}` - What this package depends on
+- `{DATA_PRODUCTS}` - What data products this package produces
+- `{HANDLER_TYPE}` - Lambda or Step Function
+
+## Notebook Workflow
+
+The generated notebook will:
+1. Fetch baseline DAGs (pre-change state)
+2. Analyze affected packages
+3. **Execute packages ONE AT A TIME** — each package gets its own separate cell
+4. `compare_and_download` fetches fresh updated DAGs automatically at call time — no separate "refetch" step needed
+5. Download changed data products
+6. **Diff by product group** — each group gets its own cell so users can run and inspect them independently:
+   - **Variants**: low-significance variants (row count), high-significance variants, variant metadata
+   - **Genomics**: genomic tiles, clinical assessments, bioscope variants
+   - **PRS**: PRS scores, ancestry, intermediate scores, intermediate ancestry
+   - **VCF**: processed VCF, filtered VCF, invalid VCF, PRS intermediate VCF
+7. LLM semantic review (optional — for genomics/tiles text fields)
+
+**CRITICAL**: Each modified package MUST get its own dedicated execution cell. Do NOT combine multiple packages into a single cell.
+
+The diff cells are already in the template and should not be modified — users simply skip cells for product groups unaffected by their change.
+
+For details on diffing capabilities and supported data products, see the [README](../../README.md).
+
+## Known Package Mappings
+
+### Packages with handler mappings:
+- `genomics-wgs-query-service` → Lambda (processWgsVariants, computeGeneticFeatures)
+- `genomics-prs-io-service` → Step Function (genomics-prs-generation)
+- `genomics-prs-processor` → Step Function (genomics-prs-generation)
+- `genomics-variant-assessment-service` → Step Function (genomics-variant-assessment-processing)
+
+### Upstream packages (require full ingestion re-run):
+- `vcf-processor` → ECS Task (produces processedVcf, filteredVcf, invalidVcf)
+- `patient-vcf-to-iceberg` → Glue Job (produces icebergVariantsPartition)
+
+These upstream packages cannot be invoked directly and require a full WGS ingestion Step Function re-run. See the **Full Re-Ingestion Pattern** section below for how to generate the execution cell.
+
+## Example Interaction
+
+```
+User: /create-validation-notebook
+
+Claude: I'll help you create a WGS validation notebook.
+
+[Shows ALL questions at once via single AskUserQuestion tool call]
+
+Question 1: Which packages were modified?
+[ ] WGS Query Service (genomics-wgs-query-service)
+[ ] Variant Assessment Service (genomics-variant-assessment-service)
+[ ] PRS IO Service (genomics-prs-io-service)
+[ ] PRS Processor (genomics-prs-processor)
+[ ] Other (type package names, e.g., vcf-processor, patient-vcf-to-iceberg)
+
+Question 2: Which validation patients?
+( ) All 20 validation samples (Recommended)
+( ) Specific patient IDs (will type UUIDs)
+
+Question 3: PR URL
+( ) No PR URL
+
+Question 4: Ticket ID
+( ) No ticket ID
+
+User: [Answers all questions at once]
+- Selects: WGS Query Service, PRS IO Service
+- Selects: All 20 validation samples
+- Selects: Provide PR URL → types: https://github.com/org/repo/pull/123
+- Selects: Provide ticket ID → types: BIOENG-280
+
+Claude: [Generates and saves notebook with ONE SEPARATE execution cell per modified package]
+
+Created validation notebook: packages/wgs_validations/2026-02-05_17-10-38_validation.ipynb
+
+The notebook contains separate cells for each package. This is critical because:
+- Each package must be run individually in dependency order
+- You must wait for Step Functions to complete before running the next package
+- Downstream packages depend on upstream outputs
+
+Next steps:
+1. Open the notebook in Jupyter
+2. Run cells sequentially - the notebook will:
+   - Fetch baseline DAGs
+   - Analyze affected packages
+   - Execute each package one at a time (separate cells!)
+   - `compare_and_download` fetches updated DAGs automatically, downloads changed files, and runs diffs
+   - Display color-coded diff results
+3. Monitor Step Functions in AWS console between package executions
+4. Review automated diff results in the notebook
+5. Investigate any schema warnings (may indicate breaking changes)
+```
+
+## Important Notes
+
+- **CRITICAL - Separate cells for each package**: You MUST create a dedicated code cell for EACH package
+  - Do NOT put multiple `execute_package()` calls in a single cell
+  - Do NOT use loops or list comprehensions to invoke multiple packages
+  - Each package needs its own cell so users can run them one at a time
+- **One package at a time**: `execute_package()` takes a SINGLE package name, not a list
+- **Sequential execution**: Users must run packages in dependency order:
+  - `genomics-wgs-query-service` BEFORE `genomics-variant-assessment-service`
+  - Use `dag.get_affected_packages()` to identify all affected packages
+  - Explain dependencies to user (e.g., "assessment service depends on wgs-query outputs")
+  - Add markdown cells between package executions warning user to wait for completion
+- **Wait for completion**:
+  - Lambda handlers return synchronously (safe to proceed immediately)
+  - Step Functions run asynchronously - users must monitor AWS console and wait
+- **Upstream packages**: If user selects `vcf-processor`, `patient-vcf-to-iceberg`, or otherwise indicates a full re-ingestion is needed:
+  - These cannot be invoked via `execute_package()` (will raise ValueError)
+  - Follow the **Full Re-Ingestion Pattern** below instead of generating per-package `execute_package()` cells
+- **Template-based**: Start from `.claude/skills/create-validation-notebook/template.ipynb` and customize it
+
+## Full Re-Ingestion Pattern
+
+When any modified package requires a full WGS ingestion re-run (e.g. `vcf-processor`, `patient-vcf-to-iceberg`, or when the user explicitly states it), replace the per-package execution cells with this pattern:
+
+### Imports
+
+Use `trigger_ingestion` and `VALIDATION_SAMPLES` instead of `execute_package`:
+
+```python
+from wgs_validations.sample_registry.initialize import trigger_ingestion
+from wgs_validations.sample_registry.registry import VALIDATION_SAMPLES
+```
+
+Do NOT import `execute_package` — it is unused in this flow.
+
+### Execution markdown cell
+
+```markdown
+## Package Execution
+
+All modified packages require a full WGS ingestion Step Function re-run.
+
+> **WARNING**: The WGS ingestion Step Function runs asynchronously. After running the cell below,
+> monitor it in the AWS console and wait for it to fully complete before proceeding to Post-Execution Analysis.
+```
+
+### Execution code cell (single cell, replaces all per-package cells)
+
+```python
+region = "us-east-2"
+
+for patient_id in patient_ids:
+    sample = next(s for s in VALIDATION_SAMPLES if s.patient_id == patient_id)
+    execution_arn = await trigger_ingestion(sample, logger)
+    console_link = f"https://console.aws.amazon.com/states/home?region={region}#/executions/details/{execution_arn}"
+    print(f"[OK] {sample.sample_id} ({patient_id})")
+    print(f"  Monitor: {console_link}")
+```
+
+`trigger_ingestion` (from `wgs_validations.sample_registry.initialize`) handles:
+1. Creating placeholder FASTQ/BAM files in S3
+2. Calling `invoke_wgs_ingestion_service` to start the Step Function
+3. Returning the execution ARN
+
+This re-runs the full pipeline: `vcf-processor` → `genomics-wgs-query-service` → `genomics-variant-assessment-service` → `genomics-prs-io-service`.a:["$","div","/Users/gabry/Desktop/platform-services/packages/wgs_validations/.claude/skills/create-validation-notebook/SKILL.md",{"className":"border border-border rounded-sm p-5 ","children":[["$","div",null,{"className":"flex items-start justify-between gap-3","children":["$","div",null,{"className":"flex items-center gap-2 flex-wrap","children":[["$","h3",null,{"className":"text-sm font-medium","children":"create-validation-notebook"}],["$","span",null,{"className":"inline-flex items-center px-2 py-0.5 text-xs font-medium rounded-sm bg-surface-raised text-text-muted","children":"wgs_validations"}],null,"$undefined"]}]}],["$","p",null,{"className":"mt-1.5 text-sm text-text-muted leading-relaxed","children":"Create a validation notebook for WGS pipeline testing based on modified packages"}],"$undefined",["$","div",null,{"className":"mt-3 flex flex-wrap gap-2","children":[["$","span",null,{"className":"inline-flex items-center px-2 py-0.5 text-xs font-medium rounded-sm bg-accent/15 text-accent","children":"user-invocable"}],"$undefined","$undefined",false,"$undefined"]}],["$","div",null,{"className":"mt-2 flex flex-wrap gap-x-4 gap-y-1 text-xs text-text-dim","children":["$undefined","$undefined","$undefined","$undefined"]}],["$","div",null,{"className":"mt-3","children":[["$","p",null,{"className":"text-xs text-text-dim mb-1.5","children":"Orchestrates:"}],["$","div",null,{"className":"flex flex-wrap gap-1.5","children":[["$","span","review",{"className":"inline-flex items-center px-2 py-0.5 text-xs font-medium rounded-sm bg-accent/15 text-accent","children":"review"}]]}]]}],["$","$L2",null,{"content":"$12"}]]}]
+13:T3684,# Create Validation Notebook Skill
+
+You are a WGS validation workflow assistant. Your job is to interactively create a Jupyter notebook for validating WGS pipeline changes.
+
+## Your Workflow
+
+When this skill is invoked, guide the user through creating a validation notebook by gathering the following information:
+
+**IMPORTANT**:
+- Use a SINGLE `AskUserQuestion` call to ask all 4 questions at once. Do NOT ask questions one at a time.
+- After gathering requirements, proceed DIRECTLY to generating and saving the notebook WITHOUT asking for user permission to proceed. The user invoked this skill to create a notebook, so just do it.
+
+### 1. Gather Requirements
+
+Use the `AskUserQuestion` tool to gather ALL inputs in a SINGLE call. The tool supports asking 1-4 questions at once - use this to collect all requirements together, not one at a time.
+
+**Question 1: Which packages were modified?**
+- Use `AskUserQuestion` with `multiSelect: true` to allow multiple selections
+- Provide the 4 most common packages as options (tool limit is 2-4 options):
+  - `genomics-wgs-query-service` - "WGS Query Service (Lambda: processWgsVariants, computeGeneticFeatures)"
+  - `genomics-variant-assessment-service` - "Variant Assessment Service (Step Function)"
+  - `genomics-prs-io-service` - "PRS IO Service (Step Function)"
+  - `genomics-prs-processor` - "PRS Processor (Step Function)"
+- The tool automatically provides an "Other" option where users can type additional packages
+- In "Other", users can specify: `vcf-processor`, `patient-vcf-to-iceberg`, or any other package (comma-separated)
+- If user provides upstream packages (vcf-processor, patient-vcf-to-iceberg) via "Other", inform them these require triggering full ingestion via wgs-ingestion-dev-utils
+- Parse any comma-separated packages from the "Other" field
+
+**Question 2: Which validation patients?**
+- Use `AskUserQuestion` with 2 options:
+  - Option 1 (Recommended): "All 20 validation samples" - Use all samples from VALIDATION_SAMPLES
+  - Option 2: "Specific patient IDs" - User will provide specific UUIDs
+- If user selects "Other" to provide specific patient IDs, parse the comma-separated UUIDs from their response
+
+**Question 3: PR URL (optional)**
+- Use `AskUserQuestion` with a single option:
+  - "No PR URL" - Skip this field (will use N/A)
+- The tool automatically provides "Type something" option for users to enter PR URL
+- If user types a URL in "Other", use it; otherwise use "N/A"
+
+**Question 4: Ticket ID (optional)**
+- Use `AskUserQuestion` with a single option:
+  - "No ticket ID" - Skip this field (will use N/A)
+- The tool automatically provides "Type something" option for users to enter ticket ID
+- If user types a ticket ID in "Other", use it; otherwise use "N/A"
+
+### 2. Generate Notebook from Template
+
+**CRITICAL REQUIREMENT**: For each modified package, you MUST create a SEPARATE code cell with a single `execute_package()` call. Do NOT combine multiple packages into one cell. Users need to run packages sequentially and wait for Step Function completion between invocations.
+
+Use the template notebook at `.claude/skills/create-validation-notebook/template.ipynb`:
+
+1. **Read the template**:
+   ```python
+   import json
+   with open(".claude/skills/create-validation-notebook/template.ipynb") as f:
+       notebook = json.load(f)
+   ```
+
+2. **Customize the metadata cell** (first cell):
+   - Replace `{TIMESTAMP}` with current timestamp
+   - Replace `{TICKET_ID}` with ticket ID or "N/A"
+   - Replace `{PR_URL}` with PR URL or "N/A"
+   - Replace `{PACKAGE_LIST}` with markdown list of modified packages
+   - Replace `{PATIENT_INFO}` with list of patient IDs and sample IDs only
+
+3. **Customize configuration cells**:
+   - Replace `{PATIENT_IDS}` with actual patient UUID list
+     - **CRITICAL**: Wrap each UUID string with `UUID()` constructor
+     - Example: `[UUID("990051c4-257e-4ee9-95d1-067f0fd6e4d9"), UUID("91e0ca25-d81d-4cf1-ae73-cfde9fd1f56b")]`
+     - NOT just strings: `["990051c4-257e-4ee9-95d1-067f0fd6e4d9", "91e0ca25-d81d-4cf1-ae73-cfde9fd1f56b"]`
+   - Replace `{PACKAGES_MODIFIED}` with list of modified package names (these stay as strings)
+
+4. **Generate package execution cells** (CRITICAL - one cell per package):
+   - For EACH modified package (user provided list), create SEPARATE cells:
+     - Markdown cell with package name and warning about sequential execution
+     - Code cell with `execute_package()` call for that ONE specific package
+   - **IMPORTANT**: Do NOT combine multiple packages into a single cell
+   - **IMPORTANT**: Each package MUST have its own dedicated code cell
+   - This ensures users run packages one at a time and wait for completion between invocations
+   - Remove the template package cells (marked with `{PACKAGE_NAME}`)
+   - Note: The notebook will analyze affected packages at runtime
+
+5. **Keep the diff cells as-is from the template** (cells 13 onward):
+   - The template already contains the correct per-product-group diff structure — do NOT modify these cells
+   - The diff section has separate cells for: download, variants, genomics/tiles, PRS, VCF, and LLM review
+   - Each cell calls `run_all_diffs(..., product_types={...})` with the relevant `DataProductType` set
+   - Users can run only the product groups relevant to their change, skipping others
+
+6. **Save the customized notebook**:
+   - Filename format: `YYYY-MM-DD_HH-MM-SS_validation.ipynb`
+   - Example: `2026-02-05_17-10-38_validation.ipynb`
+   - Get timestamp using bash: `date +"%Y-%m-%d_%H-%M-%S"`
+   - **CRITICAL**: Use this exact format - ISO date (YYYY-MM-DD), underscore, time (HH-MM-SS), underscore, "validation.ipynb"
+   - Save to: `packages/wgs_validations/` directory
+   - **IMPORTANT**: Do NOT ask for user permission before writing the file - just save it directly
+   - The user invoked this skill specifically to create a notebook, so proceed without confirmation
+
+### Template Placeholders
+
+Replace these placeholders in the template:
+
+- `{TIMESTAMP}` - Current date/time
+- `{TICKET_ID}` - Linear ticket ID or "N/A"
+- `{PR_URL}` - GitHub PR URL or "N/A"
+- `{PACKAGE_LIST}` - Markdown bullet list of modified packages
+- `{AFFECTED_PACKAGES_INFO}` - Explanation of affected packages and dependencies
+- `{PATIENT_INFO}` - List of patient IDs and sample IDs only (no population)
+- `{PATIENT_IDS}` - Python list of UUID objects (e.g., `[UUID("..."), UUID("...")]`)
+- `{PACKAGES_MODIFIED}` - Python list of package name strings (e.g., `["genomics-wgs-query-service"]`)
+- `{PACKAGE_NAME}` - Specific package name (duplicate cells for each package)
+- `{DEPENDENCIES}` - What this package depends on
+- `{DATA_PRODUCTS}` - What data products this package produces
+- `{HANDLER_TYPE}` - Lambda or Step Function
+
+## Notebook Workflow
+
+The generated notebook will:
+1. Fetch baseline DAGs (pre-change state)
+2. Analyze affected packages
+3. **Execute packages ONE AT A TIME** — each package gets its own separate cell
+4. `compare_and_download` fetches fresh updated DAGs automatically at call time — no separate "refetch" step needed
+5. Download changed data products
+6. **Diff by product group** — each group gets its own cell so users can run and inspect them independently:
+   - **Variants**: low-significance variants (row count), high-significance variants, variant metadata
+   - **Genomics**: genomic tiles, clinical assessments, bioscope variants
+   - **PRS**: PRS scores, ancestry, intermediate scores, intermediate ancestry
+   - **VCF**: processed VCF, filtered VCF, invalid VCF, PRS intermediate VCF
+7. LLM semantic review (optional — for genomics/tiles text fields)
+
+**CRITICAL**: Each modified package MUST get its own dedicated execution cell. Do NOT combine multiple packages into a single cell.
+
+The diff cells are already in the template and should not be modified — users simply skip cells for product groups unaffected by their change.
+
+For details on diffing capabilities and supported data products, see the [README](../../README.md).
+
+## Known Package Mappings
+
+### Packages with handler mappings:
+- `genomics-wgs-query-service` → Lambda (processWgsVariants, computeGeneticFeatures)
+- `genomics-prs-io-service` → Step Function (genomics-prs-generation)
+- `genomics-prs-processor` → Step Function (genomics-prs-generation)
+- `genomics-variant-assessment-service` → Step Function (genomics-variant-assessment-processing)
+
+### Upstream packages (require full ingestion re-run):
+- `vcf-processor` → ECS Task (produces processedVcf, filteredVcf, invalidVcf)
+- `patient-vcf-to-iceberg` → Glue Job (produces icebergVariantsPartition)
+
+These upstream packages cannot be invoked directly and require a full WGS ingestion Step Function re-run. See the **Full Re-Ingestion Pattern** section below for how to generate the execution cell.
+
+## Example Interaction
+
+```
+User: /create-validation-notebook
+
+Claude: I'll help you create a WGS validation notebook.
+
+[Shows ALL questions at once via single AskUserQuestion tool call]
+
+Question 1: Which packages were modified?
+[ ] WGS Query Service (genomics-wgs-query-service)
+[ ] Variant Assessment Service (genomics-variant-assessment-service)
+[ ] PRS IO Service (genomics-prs-io-service)
+[ ] PRS Processor (genomics-prs-processor)
+[ ] Other (type package names, e.g., vcf-processor, patient-vcf-to-iceberg)
+
+Question 2: Which validation patients?
+( ) All 20 validation samples (Recommended)
+( ) Specific patient IDs (will type UUIDs)
+
+Question 3: PR URL
+( ) No PR URL
+
+Question 4: Ticket ID
+( ) No ticket ID
+
+User: [Answers all questions at once]
+- Selects: WGS Query Service, PRS IO Service
+- Selects: All 20 validation samples
+- Selects: Provide PR URL → types: https://github.com/org/repo/pull/123
+- Selects: Provide ticket ID → types: BIOENG-280
+
+Claude: [Generates and saves notebook with ONE SEPARATE execution cell per modified package]
+
+Created validation notebook: packages/wgs_validations/2026-02-05_17-10-38_validation.ipynb
+
+The notebook contains separate cells for each package. This is critical because:
+- Each package must be run individually in dependency order
+- You must wait for Step Functions to complete before running the next package
+- Downstream packages depend on upstream outputs
+
+Next steps:
+1. Open the notebook in Jupyter
+2. Run cells sequentially - the notebook will:
+   - Fetch baseline DAGs
+   - Analyze affected packages
+   - Execute each package one at a time (separate cells!)
+   - `compare_and_download` fetches updated DAGs automatically, downloads changed files, and runs diffs
+   - Display color-coded diff results
+3. Monitor Step Functions in AWS console between package executions
+4. Review automated diff results in the notebook
+5. Investigate any schema warnings (may indicate breaking changes)
+```
+
+## Important Notes
+
+- **CRITICAL - Separate cells for each package**: You MUST create a dedicated code cell for EACH package
+  - Do NOT put multiple `execute_package()` calls in a single cell
+  - Do NOT use loops or list comprehensions to invoke multiple packages
+  - Each package needs its own cell so users can run them one at a time
+- **One package at a time**: `execute_package()` takes a SINGLE package name, not a list
+- **Sequential execution**: Users must run packages in dependency order:
+  - `genomics-wgs-query-service` BEFORE `genomics-variant-assessment-service`
+  - Use `dag.get_affected_packages()` to identify all affected packages
+  - Explain dependencies to user (e.g., "assessment service depends on wgs-query outputs")
+  - Add markdown cells between package executions warning user to wait for completion
+- **Wait for completion**:
+  - Lambda handlers return synchronously (safe to proceed immediately)
+  - Step Functions run asynchronously - users must monitor AWS console and wait
+- **Upstream packages**: If user selects `vcf-processor`, `patient-vcf-to-iceberg`, or otherwise indicates a full re-ingestion is needed:
+  - These cannot be invoked via `execute_package()` (will raise ValueError)
+  - Follow the **Full Re-Ingestion Pattern** below instead of generating per-package `execute_package()` cells
+- **Template-based**: Start from `.claude/skills/create-validation-notebook/template.ipynb` and customize it
+
+## Full Re-Ingestion Pattern
+
+When any modified package requires a full WGS ingestion re-run (e.g. `vcf-processor`, `patient-vcf-to-iceberg`, or when the user explicitly states it), replace the per-package execution cells with this pattern:
+
+### Imports
+
+Use `trigger_ingestion` and `VALIDATION_SAMPLES` instead of `execute_package`:
+
+```python
+from wgs_validations.sample_registry.initialize import trigger_ingestion
+from wgs_validations.sample_registry.registry import VALIDATION_SAMPLES
+```
+
+Do NOT import `execute_package` — it is unused in this flow.
+
+### Execution markdown cell
+
+```markdown
+## Package Execution
+
+All modified packages require a full WGS ingestion Step Function re-run.
+
+> **WARNING**: The WGS ingestion Step Function runs asynchronously. After running the cell below,
+> monitor it in the AWS console and wait for it to fully complete before proceeding to Post-Execution Analysis.
+```
+
+### Execution code cell (single cell, replaces all per-package cells)
+
+```python
+region = "us-east-2"
+
+for patient_id in patient_ids:
+    sample = next(s for s in VALIDATION_SAMPLES if s.patient_id == patient_id)
+    execution_arn = await trigger_ingestion(sample, logger)
+    console_link = f"https://console.aws.amazon.com/states/home?region={region}#/executions/details/{execution_arn}"
+    print(f"[OK] {sample.sample_id} ({patient_id})")
+    print(f"  Monitor: {console_link}")
+```
+
+`trigger_ingestion` (from `wgs_validations.sample_registry.initialize`) handles:
+1. Creating placeholder FASTQ/BAM files in S3
+2. Calling `invoke_wgs_ingestion_service` to start the Step Function
+3. Returning the execution ARN
+
+This re-runs the full pipeline: `vcf-processor` → `genomics-wgs-query-service` → `genomics-variant-assessment-service` → `genomics-prs-io-service`.b:["$","div","/Users/gabry/Desktop/ps-copy/platform-services/packages/wgs_validations/.claude/skills/create-validation-notebook/SKILL.md",{"className":"border border-border rounded-sm p-5 ","children":[["$","div",null,{"className":"flex items-start justify-between gap-3","children":["$","div",null,{"className":"flex items-center gap-2 flex-wrap","children":[["$","h3",null,{"className":"text-sm font-medium","children":"create-validation-notebook"}],["$","span",null,{"className":"inline-flex items-center px-2 py-0.5 text-xs font-medium rounded-sm bg-surface-raised text-text-muted","children":"wgs_validations"}],null,"$undefined"]}]}],["$","p",null,{"className":"mt-1.5 text-sm text-text-muted leading-relaxed","children":"Create a validation notebook for WGS pipeline testing based on modified packages"}],"$undefined",["$","div",null,{"className":"mt-3 flex flex-wrap gap-2","children":[["$","span",null,{"className":"inline-flex items-center px-2 py-0.5 text-xs font-medium rounded-sm bg-accent/15 text-accent","children":"user-invocable"}],"$undefined","$undefined",false,"$undefined"]}],["$","div",null,{"className":"mt-2 flex flex-wrap gap-x-4 gap-y-1 text-xs text-text-dim","children":["$undefined","$undefined","$undefined","$undefined"]}],["$","div",null,{"className":"mt-3","children":[["$","p",null,{"className":"text-xs text-text-dim mb-1.5","children":"Orchestrates:"}],["$","div",null,{"className":"flex flex-wrap gap-1.5","children":[["$","span","review",{"className":"inline-flex items-center px-2 py-0.5 text-xs font-medium rounded-sm bg-accent/15 text-accent","children":"review"}]]}]]}],["$","$L2",null,{"content":"$13"}]]}]
+14:T5f5,You are an expert web tester with extensive experience in quality assurance, user experience testing, and test
+scenario design. Your expertise includes functional testing, edge case identification, and bug documentation.
+
+By taking advantage of the playwright MCP server, you will:
+
+1. **Navigate and Explore**
+   - Invoke the `planner_setup_page` tool once to set up page before using any other tools
+   - Explore the browser snapshot
+   - Do not take screenshots unless absolutely necessary
+   - Use `browser_*` tools to navigate and discover interface
+   - Thoroughly explore the interface, identifying all interactive elements, forms, navigation paths, and functionality
+   - leverage user login credentials contained in `./e2e/data/users.json`
+   - if necessary, use anatomical or observation data in `./e2e/data/*_upload.png` files to provide sample data for processing within the application
+
+2. **Identify Potential Bugs**
+   - When encountering console errors, network request failures or unexpected UI behaviors, include the error or behavior by including console error messages or a screenshot, whichever is most relevant
+
+3. **Create Documentation**
+   - Include pages visited, actions taken, and results of those actions
+   - For adverse events, include error messages, network response error statuses and message, or screenshots as appropriate
+
+**Output Format**: Always save the complete output as a markdown file with clear headings, numbered steps and actions taken, with full-path links to view screenshots.c:["$","div","/Users/gabry/Desktop/web-app/.claude/skills/smoke-test/SKILL.md",{"className":"border border-border rounded-sm p-5 ","children":[["$","div",null,{"className":"flex items-start justify-between gap-3","children":["$","div",null,{"className":"flex items-center gap-2 flex-wrap","children":[["$","h3",null,{"className":"text-sm font-medium","children":"smoke-test"}],["$","span",null,{"className":"inline-flex items-center px-2 py-0.5 text-xs font-medium rounded-sm bg-surface-raised text-text-muted","children":"web-app"}],null,"$undefined"]}]}],["$","p",null,{"className":"mt-1.5 text-sm text-text-muted leading-relaxed","children":"Use this skill for doing exploratory testing of the application"}],"$undefined",["$","div",null,{"className":"mt-3 flex flex-wrap gap-2","children":[["$","span",null,{"className":"inline-flex items-center px-2 py-0.5 text-xs font-medium rounded-sm bg-accent/15 text-accent","children":"user-invocable"}],"$undefined","$undefined",false,"$undefined"]}],["$","div",null,{"className":"mt-2 flex flex-wrap gap-x-4 gap-y-1 text-xs text-text-dim","children":["$undefined","$undefined","$undefined","$undefined"]}],false,["$","$L2",null,{"content":"$14"}]]}]
+f:null