[codex] Move maintainer task skills under .agents (#7658)

## Thinking Path

> - Paperclip is the open source app people use to manage AI agents for
work.
> - The skills layout separates runtime Paperclip skills in `skills/`
from maintainer/agent workflow skills in `.agents/skills/`.
> - Three maintainer workflow skills still lived under root `skills/`,
making them look like runtime skills shipped through the Paperclip skill
path.
> - Root `skills/` is documented as reserved for Paperclip runtime
skills, so these task-oriented maintainer skills belong with the other
`.agents/skills` entries.
> - This pull request moves the three requested skill packages, updates
the direct smoke path, and adds regression coverage for the
maintainer-only skill boundary.
> - The benefit is a cleaner skills boundary without changing skill
contents or runtime behavior.

## Linked Issues or Issue Description

Internal Paperclip issue: PAP-10471.

No public GitHub issue exists for this repository-maintenance change.
Inline feature/enhancement issue description follows the feature request
template fields:

### Problem or motivation

Root `skills/` is documented as reserved for Paperclip runtime skills,
but `terminal-bench-loop`, `paperclip-create-plugin`, and
`diagnose-why-work-stopped` lived there even though they are
maintainer/agent workflow skills.

### Proposed solution

Move those three skill packages to `.agents/skills/`, update the
terminal-bench loop smoke script to read the new local path, and cover
the moved skill names in the existing runtime-skill discovery test
fixture.

### Alternatives considered

Leaving the skills in root `skills/` would preserve direct old paths,
but it keeps blurring the runtime-skill boundary. Moving them into the
app-shipped skills catalog would be the wrong fit because these are
maintainer workflow skills, not bundled company skills.

### Roadmap alignment

This is a small maintenance cleanup around the existing Skills
Manager/workflow-skill organization and does not introduce a
roadmap-level core feature.

## What Changed

- Moved `terminal-bench-loop`, `paperclip-create-plugin`, and
`diagnose-why-work-stopped` into `.agents/skills/`.
- Updated the terminal-bench loop smoke script and skill self-check text
to use `.agents/skills/terminal-bench-loop/SKILL.md`.
- Added regression coverage in
`server/src/__tests__/paperclip-skill-utils.test.ts` that places these
three skills under `.agents/skills` while asserting runtime discovery
still lists only root runtime skills.

## Verification

- `pnpm smoke:terminal-bench-loop-skill --source-issue-id
"$PAPERCLIP_TASK_ID" --run-key PAP-10471-move-skill-path`
- `pnpm exec vitest run
server/src/__tests__/paperclip-skill-utils.test.ts`
- `rg -n
"skills/(terminal-bench-loop|paperclip-create-plugin|diagnose-why-work-stopped)"
. --glob '!node_modules' --glob '!dist' --glob '!ui/dist'` returned no
matches.

## Risks

- Low risk: this is a file-location change plus direct path/test
updates.
- Maintainer agents that referenced the old root paths directly will
need to use `.agents/skills/...` instead.

> For core feature work, check [`ROADMAP.md`](ROADMAP.md) first and
discuss it in `#dev` before opening the PR. Feature PRs that overlap
with planned core work may need to be redirected — check the roadmap
first. See `CONTRIBUTING.md`.

## Model Used

- OpenAI Codex, GPT-5 coding agent with shell/tool use.

## Checklist

- [x] I have included a thinking path that traces from project context
to this change
- [x] I have specified the model used (with version and capability
details)
- [x] I have checked ROADMAP.md and confirmed this PR does not duplicate
planned core work
- [x] I have searched GitHub for duplicate or related PRs and linked
them above
- [x] I have either (a) linked existing issues with `Fixes: #` / `Closes
#` / `Refs #` OR (b) described the issue in-PR following the relevant
issue template
- [x] I have run tests locally and they pass
- [x] I have added or updated tests where applicable
- [x] If this change affects the UI, I have included before/after
screenshots
- [x] I have updated relevant documentation to reflect my changes
- [x] I have considered and documented any risks above
- [ ] All Paperclip CI gates are green
- [ ] Greptile is 5/5 with no open P2s, recommendations, or follow-ups
- [x] I will address all Greptile and reviewer comments before
requesting merge

.agents/skills/terminal-bench-loop/SKILL.md

@@ -228,7 +228,7 @@ The command uses the current Paperclip API token and company from `PAPERCLIP_API
 
 The smoke is deterministic and intentionally non-comparable. It does not start Terminal-Bench, Harbor, an agent model, or a provider runtime. It verifies only the control-plane shape:
 
-- local `skills/terminal-bench-loop/SKILL.md` contains the loop contract terms;
+- local `.agents/skills/terminal-bench-loop/SKILL.md` contains the loop contract terms;
 - a top-level loop issue can be created and updated into a blocker posture;
 - an iteration child issue can be created under the loop parent;
 - mocked benchmark artifact paths are recorded on a `run` document;

scripts/smoke/terminal-bench-loop-skill-smoke.mjs

@@ -83,7 +83,7 @@ function assert(condition, message) {
 }
 
 async function assertLocalSkillPackage() {
-  const skillPath = join(repoRoot, "skills", "terminal-bench-loop", "SKILL.md");
+  const skillPath = join(repoRoot, ".agents", "skills", "terminal-bench-loop", "SKILL.md");
   const markdown = await readFile(skillPath, "utf8");
   for (const expected of [
     "name: terminal-bench-loop",

server/src/__tests__/paperclip-skill-utils.test.ts

@@ -27,7 +27,10 @@ describe("paperclip skill utils", () => {
     await fs.mkdir(moduleDir, { recursive: true });
     await fs.mkdir(path.join(root, "skills", "paperclip"), { recursive: true });
     await fs.mkdir(path.join(root, "skills", "paperclip-create-agent"), { recursive: true });
+    await fs.mkdir(path.join(root, ".agents", "skills", "diagnose-why-work-stopped"), { recursive: true });
+    await fs.mkdir(path.join(root, ".agents", "skills", "paperclip-create-plugin"), { recursive: true });
     await fs.mkdir(path.join(root, ".agents", "skills", "release"), { recursive: true });
+    await fs.mkdir(path.join(root, ".agents", "skills", "terminal-bench-loop"), { recursive: true });
 
     const entries = await listPaperclipSkillEntries(moduleDir);