docs(#4025): add build commands and import conventions to boost AGENTS.md

[fullsend-ai-coder]

Add two new sections to the boost workspace AGENTS.md to prevent recurring CI failures and convention violations in agent-generated PRs:

• Build & verify: command table covering build, type-check, lint,
  test, API reports, and OpenSpec validation. Includes guidance to
  run yarn build:api-reports after modifying API surface files.
• Import conventions: documents @remixicon/react as the standard
  icon library (not @mui/icons-material) and .toLowerCase()
  normalization for spec.type comparisons.

These gaps caused preventable failures on PR #3747: a missing report.api.md regeneration, wrong icon library imports after merge, and case-sensitive specType comparisons.

---

Closes #4025

Post-script verification

• Branch is not main/master (agent/4025-boost-agents-conventions)
• Secret scan passed (gitleaks — 9cfc354095c029b95e030e58fd983bb316437f09..HEAD)
• Pre-commit hooks passed (authoritative run on runner)
• Tests ran inside sandbox

[sonarqubecloud]

Quality Gate passed

Issues
0 New issues
0 Accepted issues

Measures
0 Security Hotspots
0.0% Coverage on New Code
0.0% Duplication on New Code

See analysis details on SonarQube Cloud

[codecov]

⚠️ JUnit XML file not found

The CLI was unable to find any JUnit XML files to upload.
For more help, visit our troubleshooting guide.

[gabemontero]

interesting @rohitkrai03 @johnmcollier @mareklibra @durandom I know we've added these PR pre-submit checks in (several I think) spots in our various skill locations, but the fullsend retro / coder thinks we need to add them at the workspace specific agent level as well

is this a case of directions in higher level skills falling out of the coder agent's context?

what do you all think

[gabemontero]

interesting @rohitkrai03 @johnmcollier @mareklibra @durandom I know we've added these PR pre-submit checks in (several I think) spots in our various skill locations, but the fullsend retro / coder thinks we need to add them at the workspace specific agent level as well

is this a case of directions in higher level skills falling out of the coder agent's context?

what do you all think

yep we had a convo about that before over at https://redhat-internal.slack.com/archives/C0AR8BTMSJH/p1784229990015829 that I forgot about

[gabemontero]

interesting @rohitkrai03 @johnmcollier @mareklibra @durandom I know we've added these PR pre-submit checks in (several I think) spots in our various skill locations, but the fullsend retro / coder thinks we need to add them at the workspace specific agent level as well
is this a case of directions in higher level skills falling out of the coder agent's context?
what do you all think

yep we had a convo about that before over at https://redhat-internal.slack.com/archives/C0AR8BTMSJH/p1784229990015829 that I forgot about

and perhaps @rohitkrai03 's change at #4023 would supersede this one then

what do we think

@mareklibra @johnmcollier @durandom FYI

[gabemontero]

discussed in that slack thread ... we are going to wait on this PR and see if #4023 is sufficient for the coding agent remembering to perform the pre-submit PR checks