/forge:sprint-pro

Aliases

• /forge:sprint-pro
• /forge:sprint plan-pro (informal alias some users may type)

Purpose

This skill orchestrates optional Sprint Planning over the Stage 5 Pro Implementation Plan (agents/planner-pro.md / skills/forge-plan-pro). It performs entry verification, profile loading, persona loading, deterministic script delegation where applicable, artifact/traceability verification, and result reporting.

This skill does not perform sprint planning logic. agents/sprint-planner-pro.md is the sole authority for sprint scope selection, capacity, dependency validation, risk, allocation, traceability, validation, and completion behavior.

Stage Ownership

Sprint Planning is not a numbered pipeline stage. This skill never calls state-manager.py advance and never calls state-manager.py set. Do not duplicate or reinterpret persona logic in this skill.

Pre-flight Check

Forge Project Verification

Read pipeline/state.md:

python3 ${CLAUDE_PLUGIN_ROOT}/scripts/state-manager.py read --cwd .

Verify Forge is initialized and current_stage >= 5. If pipeline/state.md does not exist, or current_stage < 5, STOP: Sprint Planning has nothing to operate on yet. Tell the user to complete Stage 5 first (/forge:plan or /forge:plan-pro). This is a read-only check — never write to pipeline/state.md from this skill.

Do not call state-manager.py preflight --stage 6 for this check. That command validates the legacy Stage 6 prerequisite (pipeline/05-plan/task-dag.md, per references/stage-order.md), which is a different artifact shape than the Stage 5 Pro output this skill reads. Using it here would silently gate on the wrong file. Verify the Stage 5 Pro artifact directly, as described next.

Stage 5 Pro Artifact Verification

Resolve every required Stage 5 Pro input with the document resolver — never assume a flat file layout:

python3 ${CLAUDE_PLUGIN_ROOT}/scripts/read-doc.py pipeline/05-plan/implementation-plan
python3 ${CLAUDE_PLUGIN_ROOT}/scripts/read-doc.py pipeline/05-plan/work-packages
python3 ${CLAUDE_PLUGIN_ROOT}/scripts/read-doc.py pipeline/05-plan/task-breakdown
python3 ${CLAUDE_PLUGIN_ROOT}/scripts/read-doc.py pipeline/05-plan/dependency-graph
python3 ${CLAUDE_PLUGIN_ROOT}/scripts/read-doc.py pipeline/05-plan/traceability

If any required document does not resolve, STOP. List every missing artifact. Do not infer, regenerate, or substitute a missing Stage 5 Pro artifact — and do not fall back to the legacy pipeline/05-plan/task-dag.md shape from this skill. A project whose only Stage 5 output is the legacy task DAG should use the existing forge-sprint skill (/forge:sprint) instead; tell the user this explicitly rather than silently degrading.

Read pipeline/06-implementation/progress.md if it exists (optional, read-only) for done-task status. Its absence is not an error.

Load Project Profile

python3 ${CLAUDE_PLUGIN_ROOT}/scripts/load-profile.py --cwd . --stage 5 --format markdown

Sprint Planning has no dedicated stage number in references/project-type-profiles.md, so it reuses the Stage 5 profile override block — the same one Stage 5 Pro itself loads — since Sprint Planning operates directly on Stage 5's output and shares its domain (Microservices, Monolith, Library, CLI, Mobile, ML, Embedded, Infrastructure, etc.). Load and pass the result unchanged to Sprint Planner Pro. The persona's references/sprint-plan/05-workflow-governance.md Profile Overrides section controls handling of additional_artifacts, additional_steps, additional_concerns, and skip_steps, and states that replace_with does not apply to Sprint Planning. Never let a profile bypass Validation, Traceability, or the never-mutate-pipeline-state rule.

Load Sprint Planner Pro and References

Read agents/sprint-planner-pro.md, adopt Sprint Planner Pro, and follow its Reference Loading Protocol exactly. The following files are mandatory agent instructions:

references/sprint-plan/01-foundation.md
references/sprint-plan/02-capacity-dependency.md
references/sprint-plan/03-risk-allocation.md
references/sprint-plan/04-traceability-validation.md
references/sprint-plan/05-workflow-governance.md

Load each reference when the agent requires it and load all five before final validation or completion. Do not omit, summarize away, substitute, or weaken a reference instruction. Follow the Workflow in references/sprint-plan/05-workflow-governance.md exactly for plan, its Sprint Review Workflow for review, and its Retrospective Workflow alongside review.

Script Integration

scripts/sprint.py implements plan, review, and list as a deterministic, no-LLM selector, but it is hardwired to the legacy artifact shape: pipeline/05-plan/task-dag.md, T-\d+ identifiers, and pipeline/05-plan/sprint-NN.md output. It has no awareness of pipeline/05-plan/, TASK-\d+ identifiers, or the sprints/ output directory this skill uses. This skill does not modify scripts/sprint.py — the legacy forge-sprint skill continues to depend on it unchanged.

Two cases follow from this:

1. Stage 5 Pro artifacts are present (the case this skill is for). scripts/sprint.py cannot parse them. Sprint Planner Pro performs the selection itself, following the same deterministic algorithm class the script implements (topological/dependency-first ordering, carry-over tasks lead, capacity-bounded fill — see references/sprint-plan/02-capacity-dependency.md and the Workflow in references/sprint-plan/05-workflow-governance.md) but reading dependency-graph.md and task-breakdown.md directly instead of shelling out to the script. This is not a duplication of business logic for the shape the script already handles — it is the only way to support a shape the script was never built to parse, without editing the script.
2. Only a legacy task DAG is present. This skill's pre-flight fails (see Stage 5 Pro Artifact Verification) and directs the user to the existing /forge:sprint skill, which continues to call scripts/sprint.py unchanged.

Never invoke scripts/sprint.py against pipeline/05-plan/ paths — it will not find its expected files and will exit non-zero with a message about the legacy DAG, which would be confusing and incorrect in this context.

Execute

Provide Sprint Planner Pro only the context authorized by its Context Scope. Use read-doc.py for every document that may be single-file or split. Execute the requested sub-workflow:

• plan — run the persona's Workflow (steps 1–15). Produces sprint-NNN.md, sprint-NNN-capacity.md, sprint-NNN-dependencies.md, sprint-NNN-risk-register.md, sprint-NNN-traceability.md.
• review — run the persona's Sprint Review workflow. Produces sprint-NNN-review.md and sprint-NNN-metrics.md.
• retro — run the persona's Retrospective workflow (typically alongside review). Produces sprint-NNN-retrospective.md.
• list — read existing files under pipeline/05-plan/sprints/ and report sprint numbers, task counts, and done/carried/blocked status per sprint. Read-only; writes nothing.

Apply the loaded profile overrides exactly as the persona's governance rules allow.

Verification

Before reporting completion, verify through the document resolver that every plan-time (or review-time) deliverable for the affected sprint resolves successfully, and that:

• Sprint Planner Pro reports PASS for Validation and every Quality Gate.
• Every committed task resolves to existing Stage 5 lineage (WP/MOD/SPEC/REQ) — no orphan tasks, no invented lineage.
• No task is assigned to more than one owner in the sprint.
• No TASK, WP, MOD, SPEC, ADR, or any Stage 1–5 identifier was created, renamed, or removed.
• No Stage 1–5 artifact and no progress.md entry was modified.
• pipeline/state.md is unchanged from its pre-run content — diff it if there is any doubt.

If verification fails, report the failures and affected artifacts per the persona's Failure Behaviour. Do not present the sprint as usable. Do not attempt to repair a Stage 5 defect from this skill or the persona — report it upstream.

Never Mutate Pipeline State

This skill SHALL NOT call state-manager.py advance, state-manager.py set, or state-manager.py history-add. Sprint Planning has no stage number to advance to and no state field it owns. The only state-manager subcommand this skill uses is the read-only read, for the pre-flight check above.

Completion Report

Present the Sprint Planner Pro Completion Report (or, for list, the plain sprint status listing) derived from the artifacts just verified. Do not summarize the full sprint content unless requested. Explicitly state that pipeline/state.md and all Stage 1–5 artifacts are unchanged.

Next Step

Sprint Planning has no canonical next-hint — it is optional and does not gate Stage 6. After a successful plan, tell the user they may build the committed tasks directly (/forge:build), with or without running /forge:sprint-pro review at the end of the sprint. Never call state-manager.py next-hint from this skill; that command is scoped to numbered stages.

Orchestration Rules

This skill SHALL: verify Stage 5 Pro artifacts resolve before doing anything else; load the Stage 5 profile block; read Sprint Planner Pro in full; use the document resolver for every input and output; delegate to the persona's own deterministic selection logic when the Stage 5 Pro shape is present; route legacy-DAG-only projects to the existing /forge:sprint skill instead of attempting to plan against a shape it does not support; verify resolved artifacts, validation, quality gates, and traceability; and present results without ever advancing or mutating pipeline state.

This skill SHALL NOT: duplicate sprint planning logic; modify scripts/sprint.py, agents/planner.md, skills/forge-sprint/, or any existing (non--pro) sprint or plan file; redefine any Stage 1–5 artifact or TASK/WP/MOD/SPEC identifier; call state-manager.py advance, set, or next-hint; bypass a Validation or Quality Gate failure; or claim a sprint is usable when Sprint Planner Pro reports FAIL.