Skip to content

Commit de270bd

Browse files
RaduAna-Mariaclaude
andcommitted
Update CI, docs and repo scripts for the genome split
The planner and genome checker pytest jobs (Node for check-genome.mjs), the smoke and activation-gate scripts, the Studio Web flavor contract test and the genome docs follow the new skill layout. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
1 parent d594aa9 commit de270bd

11 files changed

Lines changed: 58 additions & 319 deletions

‎.claude/commands/test-coverage.md‎

Lines changed: 2 additions & 2 deletions
Original file line numberDiff line numberDiff line change
@@ -63,7 +63,7 @@ Components are the specific, testable units the skill teaches. What counts as a
6363
| RPA workflows (`uipath-rpa`) | Workflow modes (Coded C#, XAML), activity types, project types. Found in section headings like "Coded Workflows Quick Reference", "XAML Workflows Quick Reference". |
6464
| Platform operations (`uipath-platform`) | CLI command groups (`uip orchestrator`, `uip is`), API domains. Found in "CLI Overview" command tables and Task Navigation. |
6565
| Solution lifecycle (`uipath-solution`) | `uip solution` lifecycle (init, pack, publish, deploy, activate). Found in the CLI Surface Probe, Critical Rules, and Workflow sections of the SKILL.md. |
66-
| Solution design + planning (`uipath-planner`) | PDD → SDD design (Phase D: PDD analysis, architecture review, SDD generation, product/scope selection, SDD templates) plus multi-skill task derivation (Lane A/B). Found in the Critical Rules, Entry Guard, Phase D, lane summaries, and Reference Navigation sections. |
66+
| Solution design + planning (`uipath-planner`) | PDD → SDD design (`design`: PDD analysis, architecture review, SDD generation, product/scope selection, SDD templates) plus multi-skill task derivation (`plan`/`decompose`). Found in the Critical Rules, Choosing the Verb, the verb guides, and Reference Navigation sections. |
6767
| Agent development (`uipath-agents`) | Lifecycle stages (Auth, Setup, Build, Bindings, Run, Deploy), framework types (LangGraph, LlamaIndex, etc.). Found in "Lifecycle Stages" section. |
6868
| Coded apps (`uipath-coded-apps`) | Pipeline stages (Push, Pull, Pack, Publish, Deploy), app configuration concepts. Found in lifecycle and "Ship It" sections. |
6969
| Cross-cutting capability (`uipath-context-grounding`) | The **modes × surfaces** matrix from the inventory doc: BatchTransform (coded, low-code), DeepRAG (coded, low-code), Index search (coded, low-code) — 6 components from `context-grounding-patterns.md` §The Three Modes + §Surface Selection. Found in the registered inventory source(s), NOT a SKILL.md. |
@@ -240,7 +240,7 @@ Count: `(Direct + Indirect) / Total`. Report Direct and Indirect separately in t
240240

241241
**Top untested buckets (roll-up signal).** After scoring components, record the **2–3 categories with the most `None` components** as a compact, comma-separated string — this is what makes the penalty legible from the SUMMARY without opening the per-skill file. Format each as `<category> <none>/<total>`, e.g. `triggers 0/3, IS connectors 0/2, run/publish 2/8`. Pick the categories that drag the score most (largest absolute `None` count; break ties by lowest coverage %). Rules:
242242
- Derive these from the **same `None` rows** already in the Component Coverage tables — do not invent new groupings.
243-
- **Planning skills (Components N/A)** and **catalog skills with no categories**: fall back to the largest-gap **workflow-step groups** or **rule groups** instead (e.g. `Lane A 0/10, entry-guard 0/1, rules 0/5`), so the column is never empty for a scored skill.
243+
- **Planning skills (Components N/A)** and **catalog skills with no categories**: fall back to the largest-gap **workflow-step groups** or **rule groups** instead (e.g. `plan` 0/10, verb-choice 0/1, rules 0/5), so the column is never empty for a scored skill.
244244
- **Skills with no tests / planned skills:** use `all (no tests)` / `— (planned)` respectively.
245245
- Keep it under ~60 chars; it rides in one Overview cell. The full enumeration stays in the per-skill report's **Untested Features** section — this is just the headline.
246246

‎.github/workflows/smoke-skills.yml‎

Lines changed: 7 additions & 1 deletion
Original file line numberDiff line numberDiff line change
@@ -124,7 +124,13 @@ jobs:
124124
# Source of truth: the `windows` tag in each task YAML, so adding
125125
# a new Windows-only skill is just a tag, not a workflow edit.
126126
WINDOWS_TASKS=$(grep -rEl '^tags:.*\bwindows\b' tests/tasks/ 2>/dev/null | sort -u)
127-
WINDOWS_SKILLS=$(echo "$WINDOWS_TASKS" | sed -nE 's|^tests/tasks/([^/]+)/.*|\1|p' | sort -u)
127+
# A skill is skipped whole only when it owns a Windows-tagged *smoke*
128+
# task: this workflow runs `--tags smoke`, so a Windows-only e2e task
129+
# (uipath-build-from-genome's build_component) must not take the skill's Linux
130+
# smoke tasks down with it. Each Windows task file stays excluded on
131+
# its own through WINDOWS_TASKS.
132+
WINDOWS_SKILLS=$(for f in $WINDOWS_TASKS; do grep -qE '^tags:.*\bsmoke\b' "$f" && echo "$f"; done \
133+
| sed -nE 's|^tests/tasks/([^/]+)/.*|\1|p' | sort -u)
128134
129135
# Opt-in benchmarks live alongside skill tests but are NOT smoke
130136
# tests — they're large/expensive and run via the dashboard

‎.github/workflows/test-helpers.yml‎

Lines changed: 11 additions & 3 deletions
Original file line numberDiff line numberDiff line change
@@ -123,8 +123,9 @@ jobs:
123123
- name: Install pytest
124124
run: pip install pytest
125125

126+
# The genome checker tests run in the uipath-genome checker job below.
126127
- name: Run pytest
127-
run: pytest tests/tasks/uipath-planner/ -v
128+
run: pytest tests/tasks/uipath-planner/ -v --ignore=tests/tasks/uipath-planner/_shared/test_genome_check.py --ignore=tests/tasks/uipath-planner/_shared/test_genome_check_findings.py --ignore=tests/tasks/uipath-planner/_shared/test_genome_contract.py
128129

129130
pytest-agents-check:
130131
runs-on: uipath-ubuntu-latest
@@ -220,14 +221,21 @@ jobs:
220221
with:
221222
python-version: '3.13'
222223

224+
# check-genome.mjs is Node: the genome tests drive it through `node`
225+
- uses: actions/setup-node@49933ea5288caeca8642d1e84afbd3f7d6820020 # v4.4.0
226+
with:
227+
node-version: '20'
228+
223229
- name: Install pytest
224230
run: pip install pytest
225231

226232
# Pins the genome graders against the skill they grade: every worked example
227233
# passes the strict profile, the skill lists match skills/, and the execute
228-
# checks count a setting or an open item only where the skill puts it.
234+
# checks count a setting or an open item only where the skill puts it. The
235+
# contract tests hold the planner's templates and skill mapping to the
236+
# builder's checker, and each skill's section pointers to its own files.
229237
- name: Run pytest
230-
run: pytest tests/tasks/uipath-genome/_shared/ -v
238+
run: pytest tests/tasks/uipath-planner/_shared/test_genome_check.py tests/tasks/uipath-planner/_shared/test_genome_check_findings.py tests/tasks/uipath-planner/_shared/test_genome_contract.py tests/tasks/uipath-build-from-genome/_shared/ -v
231239

232240
runtime-payload-casing-guard:
233241
runs-on: uipath-ubuntu-latest

‎docs/guides/images/agents-over-time.svg‎

Lines changed: 0 additions & 17 deletions
This file was deleted.

0 commit comments

Comments
 (0)