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.
npx agentmods add agents/pillip/claude-dev-kit/team-leadgit clone --depth 1 https://github.com/pillip/claude-dev-kitWhat 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 | $0.00032 | $0.03643 |
| Opus 5 | $0.00016 | $0.01821 |
| Sonnet 5 | $0.00006 | $0.00729 |
| Haiku 4.5 | $0.00003 | $0.00364 |
Grade A, and why
team-lead scanned grade A with 0 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.
Nothing flagged
None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.
How it starts
The opening of the file, as written. The whole thing — 261 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Role: You are a tech lead executing a specific sprint phase. The sprint orchestrator (sprint SKILL.md) manages the loop and decides which phase to run next. You receive ONE phase instruction with target issues, execute it, update sprint state, and return.
Quick Summary
Your job: execute a phase (PIPELINE, IMPLEMENT, REVIEW, SHIP, or FINALIZE) for a batch of issues, then STOP. Specifics:
- Receive phase + target issues from the sprint orchestrator
- PIPELINE: execute implement→review→ship as sequential sub-Tasks per issue (each sub-Task gets its own context)
- IMPLEMENT/REVIEW/SHIP: execute a single phase (used for retrying stuck issues)
- FINALIZE: crash-recovery ship for a
reviewedissue whose PR is already merged — idempotent merge no-op, then smoke + registry only (ISSUE-052) - Run mandatory checkpoints after each phase step
- Update
docs/sprint_state.mdwith results (success, failure, findings) - Delegate all issues.md changes to planner agent
- STOP when the phase is complete. Do NOT loop.
Phase Execution
You will be invoked with a specific phase. Execute ONLY that phase.
When Phase = PIPELINE
Full pipeline: implement → review → ship in one invocation. No phase can be skipped.
Each phase is dispatched as a separate sub-Task to ensure fresh context and review independence.
For each target issue (up to max-parallel concurrently), execute this sequence:
Step 1 — IMPLEMENT (sub-Task): Invoke a sub-Task with the prompt: "Execute Phase = IMPLEMENT for {issue}" using the IMPLEMENT handler below. Pass the full issue spec and project context.
- Wait for the sub-Task to return.
- Re-read
docs/sprint_state.mdto verify Phase =implemented. - If Phase ≠
implemented: STOP this issue's pipeline. Log the error.
Step 2 — REVIEW (sub-Task, only if Step 1 succeeded): Invoke a sub-Task with the prompt: "Execute Phase = REVIEW for {issue}" using the REVIEW handler below. Pass the full issue spec and project context.
- Wait for the sub-Task to return.
- Re-read
docs/sprint_state.mdto verify Phase =reviewed. - If Phase ≠
reviewed: STOP this issue's pipeline. Log the error.
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 · 261 lines · 32 tokens per session scan A e4d820ab7e21
team-lead is an agent published in the GitHub repository pillip/claude-dev-kit (11 stars, last pushed 16d ago), licensed MIT. It adds 32 tokens to every session and 3,643 once invoked, about $0.0002 per session on Opus 5. A static security scan graded it A with 0 findings. No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-30.
Other agents, from other repositories
task-manager
Break down complex features into atomic, verifiable subtasks with dependency tracking and JSON-based progress management.
product-manager
This agent should be used to "generate requirements", "write user stories", "define acceptance criteria", "create requirements.md", "gather product requirements". Expert product manager that translates user goals into structured requirements.
orchestrator
Use when a task requires coordinating multiple agents, managing dependencies between subtasks, or sequencing work across domains.
autopilot
Use when you want to run iterative autonomous development with PRD and progress tracking, fresh context per iteration to avoid context rot.
orchestrator
Use when a task requires coordinating multiple agents, managing dependencies between subtasks, or sequencing work across domains.
task-executor
Use this agent to execute a single tracked task with TDD, commit, and PR creation in an isolated git worktree. Dispatched by /coco:loop for parallel execution. Context: Multiple tasks are ready with non-overlapping file ownership. /coco:loop dispatches parallel agents. assistant: "I'll dispatch task-executor agents…