Revert #393 — 34 lines of bash and an awk extractor to append a paragraph

[yalexx]

Reverts #393 (7c4ab60) on Yanko's call. He is right: the machinery outgrew the problem.

The problem was one sentence — the box tells customers its context is 128K when it is 1M. What #393 built for that: 34 lines of shell, an awk state machine extracting a marked section out of a template at runtime, marker-pair validation, and 11 tests covering dangling openers and orphan closers. All of that exists to append a paragraph to a markdown file without clobbering user edits — a constraint invented by the design, not by the requirement.

The 128K statement is still wrong and still worth fixing. It should come back as something smaller. The obvious candidate:

A separate managed file. CLAWBOX-AI.md, owned by us, overwritten on every gateway start, referenced from AGENTS.md the way CLAWBOX.md already is. No markers, no extractor, no idempotence question, no risk of touching anything a customer wrote — because it is not their file. Three lines of install -m 644.

Better still if the limits belong in the agent's own instructions rather than a doc it may or may not read, but that is a bigger question about where device facts live.

The two CodeRabbit findings on #393 were both real and both fixed; neither is a reason to keep the shape. Reverting now rather than iterating on top keeps beta clean while the simpler version is written.

🤖 Generated with Claude Code

Summary by CodeRabbit

• Documentation

  • Removed an unnecessary blank separator from the workspace guide.

• Behavior Changes

  • Existing CLAWBOX.md files are no longer automatically updated with the AI model limits section during gateway startup.
  • Workspace initialization and AGENTS.md reference handling remain unchanged.

• Tests

  • Removed tests covering the retired CLAWBOX.md migration behavior.

[coderabbitai]

No actionable comments were generated in the recent review. 🎉

ℹ️ Recent review info

⚙️ Run configuration

Configuration used: Path: .coderabbit.yaml

Review profile: ASSERTIVE

Plan: Pro Plus

Run ID: 5b809078-04d8-447e-af44-14f5596ea6e2

📥 Commits

Reviewing files that changed from the base of the PR and between 7c4ab60 and 02b1836.

📒 Files selected for processing (3)

• config/clawbox-workspace-guide.md
• scripts/gateway-pre-start.sh
• src/tests/unit/gateway-pre-start-clawbox-md.test.ts

💤 Files with no reviewable changes (3)

• scripts/gateway-pre-start.sh
• src/tests/unit/gateway-pre-start-clawbox-md.test.ts
• config/clawbox-workspace-guide.md

Included review availability: 0 reviews are currently available. Your included PR review attempts over the past 7 days set your current allowance at 5 reviews per hour.

---

📝 Walkthrough

Walkthrough

The gateway pre-start script no longer migrates the AI model limits section into existing CLAWBOX.md files. The related guide separator and integration test file were removed.

Changes

CLAWBOX.md migration removal

Layer / File(s)	Summary
Remove model limits migration scripts/gateway-pre-start.sh, config/clawbox-workspace-guide.md, src/tests/unit/gateway-pre-start-clawbox-md.test.ts	The startup script no longer appends the AI model limits section. The guide separator and migration integration tests were removed.

Estimated code review effort: 2 (Simple) | ~10 minutes

Merge Risk: ⚪ Minimal · up to 02b18

This change removes the runtime document-editing machinery and its associated tests; no actionable merge-blocking risk remains beyond normal checks and review.

Possibly related PRs

• ID-Robots/clawbox#82: Introduced the workspace guide seeding and migration logic reverted here.
• ID-Robots/clawbox#393: Introduced the guide section, migration logic, and tests reverted here.
• ID-Robots/clawbox#133: Also modifies scripts/gateway-pre-start.sh with a separate startup behavior.

Suggested labels: area: docs

🚥 Pre-merge checks | ✅ 4 | ❌ 1

❌ Failed checks (1 warning)

Check name	Status	Explanation	Resolution
Description check	⚠️ Warning	The description explains the rationale, but it omits the required type, testing, checklist, and structured summary sections.	Add the template sections and document the change type, test results, checklist status, and any relevant logs or screenshots.

✅ Passed checks (4 passed)

Check name	Status	Explanation
Title check	✅ Passed	The title clearly identifies the revert and accurately describes the removed shell and awk implementation.
Docstring Coverage	✅ Passed	No functions found in the changed files to evaluate docstring coverage. Skipping docstring coverage check.
Linked Issues check	✅ Passed	Check skipped because no linked issues were found for this pull request.
Out of Scope Changes check	✅ Passed	Check skipped because no linked issues were found for this pull request.

✨ Finishing Touches

📝 Generate docstrings

• Create stacked PR
• Commit on current branch

🧪 Generate unit tests (beta)

• Create PR with unit tests
• Commit unit tests in branch revert/clawbox-md-model-limits

---

_{Comment @coderabbitai help to get the list of available commands.}

[github-actions]

🦀 ClawReview

Claws waving — here's what this change is about.

Reverts #393 (7c4ab60), which added a 34-line bash block with an awk state-machine extractor and 11 tests to surgically append an AI model limits paragraph to pre-existing customer CLAWBOX.md files. Yanko called the machinery too heavy for the problem it solved, so this PR removes all of it — the shell block, the marked template section, and the full test file — leaving beta clean while a simpler follow-up approach is designed.

At a glance

• 🧹 Chore · touches gateway pre-start script, workspace guide template, unit tests
• Base branch: beta · +0 source / +0 tests across 3 files
• ✅ base beta matches the beta-first convention
• 🟡 title doesn't follow type: description (feat/fix/chore/docs/…)
• ℹ️ touches security-sensitive paths (scripts/gateway-pre-start.sh) — review with extra care

Good to know

• 🟡 The underlying problem — the box tells customers its context window is 128K when it is actually 1M — is still unresolved on customer devices after this revert. A follow-up fix is expected.
• ℹ️ Removes 152 lines of tests alongside the feature they covered; test count drops but nothing untested is left behind.
• ℹ️ gateway-pre-start.sh runs on customer hardware at every gateway start; the change is deletion-only, so no new risk is introduced on existing devices.

_{— ClawReview 🦀. I set the scene; CodeRabbit reviews the code; you decide. Conventions: docs.}

[github-actions]

CI Summary

✅ Tests

• Result: passed
• View run
• Coverage: statements 65.18%, branches 54.15%, functions 63.12%, lines 67.26%

✅ E2E

• Result: passed
• View run

✅ E2E Install

• Result: success
• View run