fix(scripts): project_render.py SKELETON_NARRATIVES matches validator (#405)

[MScottAdams]

Summary

Issue #405 reports that scripts/project_render.py SKELETON_NARRATIVES uses the TechStack camelCase key, but scripts/vbrief_validate.py PROJECT_DEF_EXPECTED_NARRATIVES requires tech stack (space-separated) after .lower(). As a result, the skeleton produced by task project:render immediately fails task vbrief:validate (part of task check).

This PR fixes the key so the skeleton passes the validator end-to-end, matching the convention already used by scripts/migrate_vbrief.py (which writes narratives["tech stack"]).

Changes

• scripts/project_render.py — rename SKELETON_NARRATIVES["TechStack"] → SKELETON_NARRATIVES["tech stack"]; added a comment block cross-referencing D3 and the validator so the contract is explicit.
• tests/cli/test_project_render.py — update the TestSkeletonCreation::test_skeleton_has_empty_narratives assertion to check the new key.
• vbrief/active/2026-04-16-405-project-render-tech-stack-key.vbrief.json — new scope vBRIEF with origin provenance to fix(scripts): project_render.py SKELETON_NARRATIVES TechStack key breaks vbrief:validate #405 and Pre-merge validation gate for vBRIEF cutover (#338) #402.
• CHANGELOG.md — Fixed entry under [Unreleased].

Why the camel-case splitter still works

_split_camel uses re.sub(r"([a-z])([A-Z])", r"\1 \2", name) and then splits on whitespace and lowercases, so _split_camel("tech stack") splits on the existing space and yields ['tech', 'stack']. Staleness flagging for the TechStack narrative is therefore unchanged by this rename.

Validation

• task check — passed locally (1539 passed, 1 xfailed).
• task project:render against this worktree emits "tech stack": "" as expected; task vbrief:validate passes on the fresh skeleton.

Scope & constraints

• Scope fence respected — only the files enumerated in fix(scripts): project_render.py SKELETON_NARRATIVES TechStack key breaks vbrief:validate #405 are touched (scripts/project_render.py, tests/cli/test_project_render.py) plus the scope vBRIEF and CHANGELOG entry.
• Other call sites that produce a TechStack narrative (run cmd_project at line 1161, documentation examples, other tests) are intentionally out of scope for this narrow fix; the cmd_project case is a downstream instance of the same bug and will be filed separately if it matters for the v0.20.0 merge gate.

Part of #402 pre-merge must-fix batch

This is PR 1 of 4 in the swarm-402 cascade on phase2/vbrief-cutover.

Closes #405
Refs #402

[greptile-apps]

Greptile Summary

Fixes the bug where SKELETON_NARRATIVES[\"TechStack\"] in scripts/project_render.py failed vbrief_validate.py's case-folded check for \"tech stack\" (\"techstack\" != \"tech stack\"), causing every fresh task project:render skeleton to immediately break task check. The rename to \"tech stack\" is correct, the _split_camel staleness-flagging helper handles the space-separated key without change, and the companion test assertion is updated in kind.

Confidence Score: 5/5

Safe to merge — targeted one-line key rename with a correct test update; no logic changes.

All findings are P2 (a stale "TechStack" mention in a docstring comment). The fix itself is mechanically correct: the renamed key case-folds to exactly what the validator expects, staleness flagging via _split_camel is unaffected, and the test assertion is updated to match.

No files require special attention.

Important Files Changed

Filename	Overview
scripts/project_render.py	Renames SKELETON_NARRATIVES["TechStack"] to "tech stack" and adds a cross-reference comment; the fix is correct and the _split_camel staleness logic remains unaffected. One stale "TechStack" reference survives in the agent-workflow docstring.
tests/cli/test_project_render.py	Updates test_skeleton_has_empty_narratives to assert "tech stack" instead of "TechStack"; all other test fixtures that use "TechStack" are testing existing-document preservation (not skeleton creation), so they remain correct.
vbrief/active/2026-04-16-405-project-render-tech-stack-key.vbrief.json	New scope vBRIEF tracking this fix; schema is valid (v0.5, status "running" in active/), origin provenance references #405 and #402, no issues.
CHANGELOG.md	Fixed entry added under [Unreleased] correctly describing the key rename, test update, and staleness-flagging continuity. No issues.

Sequence Diagram

sequenceDiagram
    participant U as User
    participant PR as project_render.py
    participant PD as PROJECT-DEFINITION.vbrief.json
    participant VV as vbrief_validate.py

    U->>PR: task project:render
    PR->>PD: create skeleton (SKELETON_NARRATIVES)
    Note over PD: narratives keys include<br/>"tech stack" (fixed)
    U->>VV: task vbrief:validate
    VV->>PD: read narratives keys
    VV->>VV: narrative_keys_lower = {k.lower() for k in narratives}<br/>check "tech stack" in narrative_keys_lower
    Note over VV: "tech stack".lower() == "tech stack" ✓<br/>(was "TechStack".lower() == "techstack" ✗)
    VV-->>U: PASS

Loading

Prompt To Fix All With AI

This is a comment left during a code review.
Path: scripts/project_render.py
Line: 25

Comment:
**Stale docstring references old key name**

The inline docstring on line 25 still says "update the `TechStack` narrative" — this should now read "`tech stack` narrative" to match the renamed key.

```suggestion
       the completed scopes (e.g. if a "tech stack" scope completed, update the
       tech stack narrative with the new technology choices).
```

How can I resolve this? If you propose a fix, please make it concise.

_{Reviews (1): Last reviewed commit: "fix(scripts): project_render.py SKELETON..." | Re-trigger Greptile}

[MScottAdams]

Acknowledging the P2 finding on scripts/project_render.py line 22-23 (stale TechStack reference in the agent-assisted layer docstring).

Not fixing in this PR per the swarm-402 charter severity ladder — P2 items are non-blocking style/nit/opinion and reply-ack is permitted. The code-level behaviour is correct and task check is green. The docstring drift will be cleaned up either as part of #410's run cmd_project pass (which also generates a TechStack narrative that will be renamed to tech stack) or as a follow-up discovered-during-402 issue.

Confidence 5/5 + all CI green + no P0/P1 ⇒ proceeding to self-merge per the PO Pre-Approval gate on #402.