[caveman] Optimize instruction verbosity — contribution-checker.agent.md (2026-05-31)

[amalgamated-bot]

Caveman Optimization Run — 2026-05-31

Applies the caveman optimization principle to instruction files:

"Why use many token when few do trick."

Files Optimized

File	Before	After	Removed
.github/agents/contribution-checker.agent.md	~111 lines	~99 lines	Redundant H1 title + description sentence; verbose comment-field prose; bloated example block

What Was Changed

• Removed the H1 title and opening description — both were verbatim restatements of the frontmatter description field.
• Tightened the Comment Field numbered list — each item shortened while preserving the required signal (opening, actionable feedback, prompt block).
• Compressed the needs-work example — trimmed verbose intro sentence, shortened bullet text, and collapsed the multi-line prompt block into one line. The structure (greeting, bullets, fenced prompt) is fully preserved.

What Was Preserved

All field names, verdict rules, checklist items, JSON schema, output format constraints, and security rules were kept intact.

Round-Robin Progress

Processed files 1–5 of 7 total files in the queue (wrapping from index 6 → 0–4).

Warning

Firewall blocked 1 domain

The following domain was blocked by the firewall during workflow execution:

• localhost

To allow these domains, add them to the network.allowed list in your workflow frontmatter:

network:
  allowed:
    - defaults
    - "localhost"

See Network Configuration for more information.

Generated by Daily Caveman Optimizer · sonnet46 1.1M · ◷

Add this agentic workflows to your repo

To install this agentic workflow, run

gh aw add github/gh-aw/.github/workflows/daily-caveman-optimizer.md@657ab8db03c70cf22bf8cbe7f6d05af8096e990b

• expires on Jun 3, 2026, 4:49 AM UTC

Greptile Summary

This PR compresses the contribution-checker.agent.md instruction file by removing the redundant H1 heading and opening description sentence, tightening the three comment-field rules, and condensing the needs-work example block. All verdict rules, checklist items, JSON schema, and output-format constraints are preserved intact.

• Removed: H1 title and its description sentence (both duplicated the frontmatter description field).
• Tightened: Comment-field items 1–3 — shorter phrasing, plus the lgtm omit-prompt-block rule folded into item 3.
• Condensed example: The multi-line needs-work prompt block collapsed to a single line; bullet text shortened while keeping the same structure.

Confidence Score: 4/5

Safe to merge — changes are limited to verbal compression of an agent instruction file with no functional logic removed.

The removal of 'Be constructive and specific' and 'tied to the checklist results' from item 2 is a minor reduction in quality-shaping guidance that could produce slightly less grounded feedback in practice, but the core evaluation flow is entirely unchanged.

.github/agents/contribution-checker.agent.md — confirm the dropped quality-guidance phrases in item 2 are acceptable.

Important Files Changed

Filename	Overview
.github/agents/contribution-checker.agent.md	Verbal compression of agent instruction file — removes H1 title/description, tightens comment-field rules, and condenses the needs-work example. Core logic, JSON schema, verdict rules, and checklist items are all preserved; minor quality-guidance phrases were dropped.

Flowchart

%%{init: {'theme': 'neutral'}}%%
flowchart TD
    A[Start: PR Input] --> B{CONTRIBUTING.md\nexists?}
    B -- No --> C[Return ❓ no-guidelines]
    B -- Yes --> D[Step 2: Gather PR Data\nnumber, title, body, diff, files]
    D --> E[Step 2.5: Targeted Context\nread linked issue if any]
    E --> F[Step 3: Run Checklist\non-topic, process, focused,\ndeps, tests, description, diff-size]
    F --> G{Step 4: Verdict Rules}
    G -- on_topic=no OR\nprocess violation --> H[🔴 Off-Guidelines]
    G -- focused=no --> I[⚠️ Needs Focus]
    G -- new deps OR\nunclear/discussion --> J[🟡 Needs Discussion]
    G -- none triggered --> K[🟢 Aligned]
    H & I & J & K --> L[Step 5: Quality Signal]
    L -- 🔴 + no desc\n+ no purpose --> M[spam]
    L -- ⚠️/🟡 OR\nmissing tests/desc --> N[needs-work]
    L -- 🟢 + tests\n+ description --> O[lgtm]
    M & N & O --> P[Output JSON\nwith comment field]
    P --> Q{quality?}
    Q -- lgtm --> R[comment: congratulate\nno prompt block]
    Q -- needs-work/spam --> S[comment: opening +\nfeedback + prompt block]

Loading

Prompt To Fix All With AI

Fix the following 1 code review issue. Work through them one at a time, proposing concise fixes.

---

### Issue 1 of 1
.github/agents/contribution-checker.agent.md:80
The phrase "Be constructive and specific" was dropped from item 2 along with "tied to the checklist results." Both serve as quality-shaping signals for the agent — without them, feedback suggestions in the `comment` field may become generic rather than grounded in the actual checklist outcomes. Re-adding just "Be specific" preserves the intent without adding tokens back.

```suggestion
2. **Actionable feedback** — for `needs-work` / 🟡 / ⚠️ / 🔴, list concrete suggestions tied to checklist results (missing tests, unfocused diff, missing description). Be specific.
```

_{Reviews (1): Last reviewed commit: "chore: caveman optimize contribution-che..." | Re-trigger Greptile}

Greptile also left 1 inline comment on this PR.

[Copilot]

Pull request overview

This PR streamlines the contribution checker agent instructions by removing duplicated introductory text and tightening the comment guidance/examples while preserving the agent’s evaluation rules and output schema.

Changes:

• Removed redundant title/description already covered by frontmatter.
• Condensed comment-field requirements and needs-work example wording.
• Preserved verdict rules, checklist fields, JSON schema, and safety constraints.

[greptile-apps]

The phrase "Be constructive and specific" was dropped from item 2 along with "tied to the checklist results." Both serve as quality-shaping signals for the agent — without them, feedback suggestions in the comment field may become generic rather than grounded in the actual checklist outcomes. Re-adding just "Be specific" preserves the intent without adding tokens back.

Suggested change

	2. **Actionable feedback** — for `needs-work` / 🟡 / ⚠️ / 🔴, list concrete suggestions (missing tests, unfocused diff, missing description).
	2. **Actionable feedback** — for `needs-work` / 🟡 / ⚠️ / 🔴, list concrete suggestions tied to checklist results (missing tests, unfocused diff, missing description). Be specific.

Prompt To Fix With AI

This is a comment left during a code review.
Path: .github/agents/contribution-checker.agent.md
Line: 80

Comment:
The phrase "Be constructive and specific" was dropped from item 2 along with "tied to the checklist results." Both serve as quality-shaping signals for the agent — without them, feedback suggestions in the `comment` field may become generic rather than grounded in the actual checklist outcomes. Re-adding just "Be specific" preserves the intent without adding tokens back.

```suggestion
2. **Actionable feedback** — for `needs-work` / 🟡 / ⚠️ / 🔴, list concrete suggestions tied to checklist results (missing tests, unfocused diff, missing description). Be specific.
```

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

Note: If this suggestion doesn't match your team's coding style, reply to this and let me know. I'll remember it for next time!

[amalgamated-bot]

Hey @amalgamated-bot 👋 — thanks for the automated token-trimming pass on contribution-checker.agent.md! Keeping agent instruction files lean is a worthwhile goal. There are a couple of things that need to be addressed before this can move forward:

• PR title doesn't follow Conventional Commits format — the CONTRIBUTING.md requires PR titles to use the Conventional Commits convention (e.g. docs: optimize contribution-checker agent instructions). The current [caveman] Optimize instruction verbosity — ... prefix isn't recognised by the project's tooling and may break changelog generation. Please rename to something like docs: trim verbosity in contribution-checker.agent.md.
• No tests — while this is a docs/agent-instructions change, if any behaviour-affecting prompt text was altered it would be worth noting explicitly what was validated. If the project has any prompt-evaluation or agent integration tests, touching them (or at least confirming they still pass) would strengthen confidence in the change.
• No linked issue / prior discussion — the CONTRIBUTING.md asks contributors to open an issue first for feature/improvement work. Linking an issue (or briefly noting that this is a routine automated maintenance task that doesn't require one) would help maintainers triage quickly.

Rename the PR title to follow Conventional Commits format. A suitable title would be:

  docs: trim verbosity in contribution-checker.agent.md

Also add a one-line note to the PR body clarifying whether an issue is required for this type of automated maintenance PR, per the project's CONTRIBUTING.md contribution process.

Warning

Firewall blocked 1 domain

The following domain was blocked by the firewall during workflow execution:

• localhost

To allow these domains, add them to the network.allowed list in your workflow frontmatter:

network:
  allowed:
    - defaults
    - "localhost"

See Network Configuration for more information.

Generated by Contribution Check · sonnet46 2.2M · ◷

Add this agentic workflows to your repo

To install this agentic workflow, run

gh aw add github/gh-aw/.github/workflows/contribution-check.md@2b0348a9b4c50a71de650ec9a5b865c385a9d0df