microsoft/hve-core

Commit

feat(graphify): Add graphify skill with thin prompt + applyTo instructions (#1600)

This PR replaces #1518 based on the feedback from Katrien De Graeve.

## Summary

Reshape the graphify contribution per reviewer feedback. Drops the
skill, agent, Python wrapper, fuzz harness, corpus, `pyproject.toml`,
and pinned `graphifyy==0.5.4` dependency. Upstream `graphifyy` already
ships first-class VS Code Copilot Chat integration via `graphify vscode
install` — hve-core has no integration gap to fill there.

This PR keeps the value (audit-tagged reporting discipline, RPI
integration, sensitive-tree guidance) while removing the cost (Python
dependency pinning against a fast-iterating personal project,
fuzz-harness surface area, and the endorsement footprint of a
Microsoft-branded skill wrapping an unaffiliated upstream).

## What hve-core uniquely adds

- **`/graph-research` slash prompt** (`agent: Task Researcher`) encoding
house conventions for consuming graph evidence inside the RPI workflow:
  - graph-beats-grep / grep-beats-graph routing
  - audit-tag reporting (`EXTRACTED` / `INFERRED` / `AMBIGUOUS`)
  - sensitive-tree fallback to `--mode fast`
  - never triggers a graph rebuild
- **`graphify.instructions.md`** scoped to `applyTo:
'**/graphify-out/**'` so audit-tag discipline, upload guidance
(Anthropic and Moonshot AI data-residency caveats), and rebuild-cost
discipline auto-apply whenever Copilot reads files under any
`graphify-out/` directory.

## Changes

- Added
[.github/prompts/experimental/graph-research.prompt.md](.github/prompts/experimental/graph-research.prompt.md)
- Added
[.github/instructions/experimental/graphify.instructions.md](.github/instructions/experimental/graphify.instructions.md)
- Wired both artifacts into the `experimental` and `hve-core-all`
collections
- Regenerated plugin symlinks and READMEs via `npm run plugin:generate`
- Added `graphify` and `graphifyy` to the spell-check dictionary

## Files

| Status | Path |
| ------ | ---- |
| A |
[.github/prompts/experimental/graph-research.prompt.md](.github/prompts/experimental/graph-research.prompt.md)
|
| A |
[.github/instructions/experimental/graphify.instructions.md](.github/instructions/experimental/graphify.instructions.md)
|
| M |
[collections/experimental.collection.yml](collections/experimental.collection.yml)
|
| M |
[collections/experimental.collection.md](collections/experimental.collection.md)
|
| M |
[collections/hve-core-all.collection.yml](collections/hve-core-all.collection.yml)
|
| M |
[collections/hve-core-all.collection.md](collections/hve-core-all.collection.md)
|
| M | [.cspell.json](.cspell.json) |
| A/M | `plugins/**` (generated) |

## Validation

- `npm run plugin:generate`
- `npm run plugin:validate`
- `npm run lint:all`

## Notes for reviewers

- No Python dependency or skill package is introduced; no
`pyproject.toml`, fuzz harness, or pinned upstream.
- The `applyTo` glob restricts the instructions to `graphify-out/` trees
so it does not leak into unrelated contexts.
- `plugins/**` changes are generated — review the source under
`.github/` and `collections/` and confirm plugin outputs match.

---------

Signed-off-by: dependabot[bot] <support@github.com>
Co-authored-by: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Co-authored-by: dependabot[bot] <49699333+dependabot[bot]@users.noreply.github.com>
Co-authored-by: Marcel Bindseil <marcelbindseil@gmail.com>
Co-authored-by: Bill Berry <WilliamBerryiii@users.noreply.github.com>