[Chennai] Vinodha — Vibe Coding Submission

[Vinodha-23]

RAG-to-MCP — Submission PR

Name: Vinodha
Email: vinodha235@gmail.com
City / Group: Chennai
Date: 2026-08-16
AI tool(s) used: Antigravity

---

Submission Checklist

• uc-0a/agents.md — present and updated
• uc-0a/skills.md — present and updated
• uc-0a/classifier.py — runs without crash
• uc-0a/results_[city].csv — output present
• uc-rag/agents.md — present and updated
• uc-rag/skills.md — present and updated
• uc-rag/rag_server.py — not the stub, your implementation
• uc-mcp/agents.md — present and updated
• uc-mcp/skills.md — present and updated
• uc-mcp/mcp_server.py — passes at least one test_client.py test
• 3+ commits with meaningful messages, one per UC
• All sections below filled

---

UC-0A — Complaint Classifier

Which failure mode did you encounter first?

Severity blindness. The AI was missing urgent complaints like children falling or injuries because the standard instructions weren't explicit enough about those keywords.

Which enforcement rule fixed it? Quote from your agents.md:

"Priority must be Urgent if description contains any severity keyword: injury, child, school, hospital, ambulance, fire, hazard, fell, collapse."

Your commit message for UC-0A:

UC-0A Fix severity blindness: no keywords in enforcement -> added injury/child/school/hospital triggers

Verification checkpoints:

• All severity-signal rows (injury/child/school/hospital keywords) classified as Urgent
• No invented categories outside the defined taxonomy
• Justification column present and non-empty for every row

---

UC-RAG — RAG Server

Which failure mode did you encounter?
(chunk boundary / wrong retrieval / answer outside context)

Answer outside context (and wrong retrieval due to vague boundaries initially).

What chunking strategy did you use and why?

I used a sentence-boundary aware chunking strategy (maximum 400 tokens per chunk but never splitting mid-sentence) so that critical clauses are kept intact and not split ambiguously.

Did your system correctly refuse "What is the flexible working culture?"?
(Should return refusal template — not in any document)

Yes

Did your system retrieve the correct document for "Can I use my personal phone for work files?"?
(Should retrieve IT policy, not HR leave policy)

Yes

Which enforcement rule in agents.md prevented answers outside retrieved context?

"Answer must use only information present in the retrieved chunks. Never add context from outside the retrieved set."

Your commit message for UC-RAG:

UC-RAG Fix context breach: no grounding enforcement -> added retrieved-chunks-only rule

Verification checkpoints:

• At least 3 test queries return grounded answers (cited from retrieved context)
• "What is the flexible working culture?" returns the refusal template (not a hallucinated answer)
• "Can I use my personal phone for work files?" retrieves IT policy, not HR leave policy
• Chunking produces more than 1 chunk per document (not whole-document embedding)

---

UC-MCP — MCP Server

Paste your tool description from mcp_server.py TOOL_DEFINITION:

"Query CMC HR Leave Policy, IT Acceptable Use Policy, and Finance Reimbursement Policy. Questions outside these three documents will be refused."

Does it state the document scope explicitly?

Yes

Run result: python3 test_client.py --run-all
(Paste the summary output)

============================================================
Reference verification complete.
Review ⚠️ warnings above — they indicate enforcement gaps.

Did the budget forecast question return isError: true?

Yes

In one sentence — why is the tool description the enforcement?

Because agents read the tool description to decide if they should call it; an explicit scope stops them from making out-of-scope calls.

Your commit message for UC-MCP:

UC-MCP Fix vague tool description: no scope stated -> added CMC policy scope + refusal note

Verification checkpoints:

• Tool description explicitly states document scope (which policies are covered)
• Tool description states refusal behavior for out-of-scope queries
• python3 test_client.py --run-all executes without connection error
• Budget forecast question returns isError: true (out of scope)

---

CRAFT Reflection

Which step of the CRAFT loop was hardest across all three UCs?

Testing the failure modes (the T in CRAFT) is the hardest because sometimes the AI correctly guesses the answer despite a bad prompt, hiding the enforcement gap until a real edge case hits.

What did you add to agents.md manually that the AI did not generate?

I had to manually tighten the severity keywords and add strict refusal templates, as the AI initially generated polite but loose boundaries.

One specific task in your real work where you will use R.I.C.E in the next 7 days:

I will use R.I.C.E to generate structured project summary reports, specifically enforcing output schemas and tone.

[github-actions]

Hi there, participant! Thanks for joining our RAG-to-MCP Workshop!

We're reviewing your PR for the 3 Use Cases (UC-0A, UC-RAG, UC-MCP). Once your submission is validated and merged, you'll be awarded your completion badge!

Next Steps:

• Make sure all 3 UCs are finished.
• Ensure your commit messages match the required format.
• Fill out every section of the PR template.
• Good luck!

[github-actions]

Hi there, participant! Thanks for joining our RAG-to-MCP Workshop!

We're reviewing your PR for the 3 Use Cases (UC-0A, UC-RAG, UC-MCP). Once your submission is validated and merged, you'll be awarded your completion badge!

Next Steps:

• Make sure all 3 UCs are finished.
• Ensure your commit messages match the required format.
• Fill out every section of the PR template.
• Good luck!