[Control Plane] Add tests for handlers/agentic — query + batch + discover

[santoshkumarradha]

Test coverage: Agentic handlers — query.go + batch.go + discover.go

control-plane/internal/handlers/agentic/ has only agentic_test.go and status_test.go. The main endpoints (query, batch, discover) are untested.

Scope

New test files:

• control-plane/internal/handlers/agentic/query_test.go
• control-plane/internal/handlers/agentic/batch_test.go
• control-plane/internal/handlers/agentic/discover_test.go

Test cases

query.go — AgenticQueryHandler / related helpers:

• Empty query string → 400 with descriptive error
• Valid query → dispatches to reasoner, returns agentic response
• Upstream reasoner error → 502 with source information
• Unknown agent id → 404
• stream=true → streaming response (SSE) instead of JSON
• Context cancellation mid-stream → handler unwinds cleanly

batch.go — AgenticBatchHandler:

• Empty batch → 400
• Partial success: one item fails, others succeed → 207-style response with per-item status
• All success → 200 with all results
• Malformed item in batch → item-level error, other items still processed

discover.go — AgenticDiscoverHandler:

• Empty filters → full list returned (up to default limit)
• Tag filter → only matching agents
• Limit + offset pagination works
• format=xml path returns correct content-type

Fixture / mock boundary

Reuse patterns from agentic_test.go and status_test.go. Stub storage + stub upstream reasoner via httptest.NewServer for query/batch.

Acceptance criteria

• Three new test files
• go test ./control-plane/internal/handlers/agentic/... passes
• Package coverage jumps

Size: L · Difficulty: intermediate · Language: Go

---

Part of epic #387 — parent tracker.

---

How to work on this

Setup

git clone https://github.com/Agent-Field/agentfield.git
cd agentfield
make install        # installs Go, Python (via uv), and Node deps
make build          # optional — builds the control plane binary

Run the baseline test suite

Before you change anything, run the tests for the language this issue targets so you know what "green" looks like:

• Go control plane: cd control-plane && go test ./... -count=1
• Go SDK: cd sdk/go && go test ./... -count=1
• Python SDK: cd sdk/python && uv run pytest -q
• TypeScript SDK: cd sdk/typescript && bun test

Package coverage baseline (so you can show improvement in the PR):

• go test -cover ./control-plane/internal/...
• uv run pytest --cov=agentfield sdk/python/tests
• bun test --coverage (TypeScript)

Using Claude Code or another AI coding agent

This repo has a CLAUDE.md at the root that Claude Code and compatible agents will read automatically. A working flow:

1. cd into the repo and run claude (or open in your IDE's agent).
2. Paste this issue URL and say: "Please implement this issue, following every acceptance-criteria checkbox. Do not change source files — only add tests under the target path in the issue. When you're done, run the test suite command from the issue and paste the result."
3. Let the agent read existing tests in the same package as a model for style / fixtures — we point to them in the issue.
4. Run make lint before opening the PR.

See the AI-Assisted Contributions guide for testing requirements. Tag your PR with ai-assisted if you used an AI agent.

Claiming this issue

Comment I'd like to work on this and a maintainer will assign it to you. If you're new to open source, issues labelled good first issue or first-timers-only are hand-picked for a smooth first PR — we'll guide you if you get stuck.

Opening the PR

• Branch name: tests/<short-description> or refactor/<short-description>
• Link this issue in the PR description with Closes #<number>
• Paste the before/after test output or coverage delta in the PR body
• Keep the diff to the scope in this issue — no drive-by cleanups

Help & questions

• Comment on this issue — we monitor all assigned issues.
• Join the community Discord: run /gsd-join-discord or check the README for the invite.
• For repo-wide questions, open a discussion with the help tag.