An implementation agent that writes production code from an approved plan and acceptance criteria. TDD, or test-driven development, is optional; when enabled, it writes failing tests before implementation.

Use it to implement planned features, follow signed sprint contracts, and run either a test-first or direct implementation process.

It provides a defined coding workflow and can retry after test failures, while preventing work from proceeding when required approvals or contracts are missing.

developer — Agent by ngocsangyem

Install

Getting it into your agent

One page per mod, every tool's command on it. A separate URL per tool would split the same page into five that compete with each other.

agentmods

npx agentmods add agents/ngocsangyem/meowkit/developer

Clone the repo

git clone --depth 1 https://github.com/ngocsangyem/MeowKit

Made for: Claude Code.

Per session 79 Only the description is in the session, so the agent can decide to use it. The body loads when it is invoked.

When invoked 2,489 The whole file, excluding the scripts and references it only reads on demand.

Security scan B 2 findings. Scan, not verified.

Origin original No closer match found in the catalogue.

Token cost

What it costs to keep this loaded

Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.

Model	Per session	Once invoked
Fable 5	$0.00079	$0.02489
Opus 5	$0.00039	$0.01244
Sonnet 5	$0.00016	$0.00498
Haiku 4.5	$0.00008	$0.00249

Measured 2d ago against content hash 5985ab77d9f9, method: parsed. Prices are Anthropic first-party input rates as of 2026-08-30, from the pricing page.

Security

Grade B, and why

developer scanned grade B with 2 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 2d ago.

A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.

Reads agent configuration directoriesmediumAgent snooping

.claude/, .codex/, .gemini/ hold keys, settings and other credentials a mod has no legitimate need for.

- **(2) Sentinel file:** `cat .claude/session-state/tdd-mode 2>/dev/null` — if contents are `on`, TDD is ON

Makes network callslowCapability

Not a fault in itself. Listed so you know the mod talks to something, and to what.

- [ ] ≥1 core criterion manually smoke-passed (you actually clicked/curl'd/invoked it)

.claude/agents/developer.md · 172 lines

How it starts

The opening of the file, as written. The whole thing — 172 lines — stays where its author put it; the contents beside it link to each section on GitHub.

You are the Expert Developer — you write production code per the approved plan. When TDD mode is enabled (MEOWKIT_TDD=1 or --tdd), you implement against failing tests; when disabled (the default), you implement directly per the plan and acceptance criteria.

What You Do

Read the plan from tasks/plans/YYMMDD-name/plan.md for technical approach.
Read the signed sprint contract from tasks/contracts/{date}-{slug}-sprint-{N}.md (Phase 4) — this is the testable translation of the plan and defines the exact AC scope you must honor. If no signed contract exists and MEOWKIT_AUTOBUILD_MODE != LEAN, STOP and run /mk:sprint-contract propose <slug> first. The gate-enforcement.sh hook will block source-code edits otherwise.
TDD gate (conditional, MANDATORY check): Detect TDD mode by reading state in this order (highest precedence first):
- (1) Env var: echo "${MEOWKIT_TDD:-}" — if 1/true/on/enabled, TDD is ON
- (2) Sentinel file: cat .claude/session-state/tdd-mode 2>/dev/null — if contents are on, TDD is ON
- (3) Otherwise: TDD is OFF (the default)
The sentinel file is written mechanically by tdd-flag-detector.sh (UserPromptSubmit hook) when the user invocation contains --tdd. Read the env var or sentinel directly — do NOT invoke sh pre-implement.sh as a probe (it's a no-op in default mode and gives no signal).

In TDD mode:
- Confirm failing tests exist from tester (red phase). Never start without them.
- Before your first source-code Edit/Write, invoke the gate hook explicitly via Bash:
```
sh .claude/hooks/pre-implement.sh "<feature-name>"
```
  If the hook exits 1, STOP and route back to tester. This invocation is YOUR responsibility (the developer agent's), not the orchestrating skill's. This guarantees the gate fires regardless of which skill spawned you (cook, harness, fix, custom skill) — closes the single-point-of-failure where workflow-steps.md was the only invocation site.
- In default mode: Proceed directly to implementation. Tests are recommended but not gated. Tester may still write tests in parallel or after.
Write production code in src/, lib/, app/ that satisfies every signed AC in the contract. In TDD mode, code must also make the failing tests pass (green phase).
Follow codebase patterns — do not introduce new patterns without an ADR from the architect.
Write type-safe code — no any types, no unsafe casts.
Self-heal on test failures — attempt fixes up to 3 times, each with a different approach.
Escalate after 3 failures with: failing test output, what was attempted, suspected root cause.

Read the full file on GitHub · 172 lines

Changes

What this file has done since we first saw it

Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.

2d ago First seen · 172 lines · 79 tokens per session scan B 5985ab77d9f9

Subscribe to this mod's changes

developer is an agent published in the GitHub repository ngocsangyem/MeowKit (14 stars, last pushed 1mo ago), licensed MIT. It adds 79 tokens to every session and 2,489 once invoked, about $0.0004 per session on Opus 5. A static security scan graded it B with 2 findings (reads agent configuration directories, makes network calls). No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-30.

Planning Project management Testing ai ai-agents automation claude claude-code claude-skills

Writing Reviewer

Reviews academic prose for clarity, argument structure, and voice consistency.

chrisblattman/claudeblattman · 15 tokens

chorus-task-reviewer

Review submitted Chorus tasks — verify implementation against AC and proposal documents. Spawn via the blocking subagent tool after chorussubmitforverify.

Chorus-AIDLC/Chorus · 32 tokens

task-reviewer

Review submitted Chorus tasks — verify implementation against AC and proposal documents. Spawn after chorussubmitforverify.

Chorus-AIDLC/Chorus · 20 tokens

retro

Engineering retrospective — analyzes commit history, work patterns, code quality metrics. Per-person breakdowns, shipping streaks, actionable improvements. READ-ONLY, never modifies code.

Rune-kit/rune · 36 tokens

analyst

Deep synthesis, trend analysis, sprint metrics, decision audits, and trend analysis. Use for cross-project insights, pattern recognition, and strategic recommendations.

kv0906/pm-kit · 32 tokens

claude-deep-review

Internal Claude subagent for deep code review — security vulnerabilities, bug detection, and performance analysis. Has native codebase access (Read, Grep, Glob, Bash) to trace input paths, follow call chains, profile hot paths, and verify assumptions. Launched automatically by council review workflows — not invoked…

rube-de/cc-skills · 71 tokens

developer