Agent Harness Sprint Plan

---
name: agent-harness-sprint-plan
description: Use when a user wants a Codex-ready plan for an agent-harness sprint before implementation. Produces a decision-complete Planner/Generator/Evaluator sprint plan without editing application code.
argument-hint: "[product spec or implementation goal]"
---

# Agent Harness Sprint Plan

Plan a Codex sprint without implementing it. Use this skill for ambiguous or multi-step work where a plan should be reviewed before edits.

## References

- `../../references/codex-sprint-contract.md` - Codex sprint artifact contract and task rules
- `../../references/codex-config-schema.md` - Codex model routing config

## Step 1 - Resolve Codex Model Routing

Read Codex config in this order:

1. `./.codex/agent-harness.local.json`
2. `~/.codex/agent-harness.json`
3. Built-in defaults from `../../references/codex-config-schema.md`

Only read `.codex` config files. Do not read `.claude/agent-harness*.json`.

If no config exists, use the built-in default where every role has
`mode: "inherit"`.

Resolve each role independently. In generated plans, state for each task
whether its assigned role:

- inherits the current Codex session model and reasoning, or
- uses an explicit `model` and optional `reasoning_effort`

If a route is malformed, warn and treat only that role as `inherit`.

## Step 2 - Ground the Plan

Read only the files needed to understand the request. Prefer `rg` and targeted file reads.

Identify:

- Goal and user-facing outcome
- Affected subsystem or files
- Existing conventions to preserve
- Risks, dependencies, and likely verification commands
- Which tasks are best mapped to `planner`, `evaluator`, `code`, `write`,
  `research`, or `collect` routing roles

Do not edit files during this skill.

## Step 3 - Write the Sprint Plan

Produce a plan in the format from `../../references/codex-sprint-contract.md`.

The plan must include:

- Interpretation
- Acceptance Criteria
- parallel_batch
- sequential_tasks
- Verification Plan
- Assumptions

Use 3 to 7 tasks. A task belongs in `parallel_batch` only when its write ownership is disjoint from every other parallel task. Read-heavy exploration, test analysis, and documentation review may run in parallel.

Each task should state the intended routing role or an equivalent note in its
summary or deliverables when that affects execution quality or cost.

## Step 4 - Codex Delegation Notes

Add explicit instructions for future execution:

- Which tasks should use parallel subagents
- Which tasks must stay sequential
- Which files or modules each worker owns
- Which tasks are read-only
- Whether each subagent should inherit the current Codex session model or use
  an explicit `model` and optional `reasoning_effort`

Codex only spawns subagents when the user or skill explicitly asks for parallel delegation. State that execution should use subagents only for the listed parallel tasks.

## Output

Return the plan to the user. If the user asks to persist it, write it under `.sprint/<YYYYMMDD-HHmmss>/sprint-plan.md` and create only the sprint artifact directory needed for that plan.