From project-management
Review and score Statement of Work documents across 8 quality dimensions with adversarial analysis. Produces a structured scorecard (1-5 per dimension), critical issues list, improvement recommendations, and adversarial challenges from both client and vendor perspectives. Use this skill whenever the user wants to review a SOW, check SOW quality, do an adversarial review of a statement of work, validate a proposal, audit a capitolato tecnico, or mentions 'revisiona SoW', 'SOW quality check', or 'review proposal document'.
How this skill is triggered — by the user, by Claude, or both
Slash command
/project-management:sow-reviewThe summary Claude sees in its skill listing — used to decide when to auto-load this skill
This skill reviews Statement of Work documents against 8 quality dimensions, producing a structured scorecard with actionable recommendations. It includes an adversarial pass that role-plays as both client and vendor to identify exploitable gaps.
This skill reviews Statement of Work documents against 8 quality dimensions, producing a structured scorecard with actionable recommendations. It includes an adversarial pass that role-plays as both client and vendor to identify exploitable gaps.
Inputs:
Connector support:
Read the SOW and all addenda. Determine:
If corporate standards are provided (or available via ~~knowledge base), load them for comparison.
Activation criteria: the user explicitly requests domain-specific validation, OR the SOW covers a regulated domain (public procurement, healthcare, financial services) where industry-standard requirements apply.
If activated, dispatch parallel subagents to research relevant standards:
Each subagent receives relevant SOW excerpts (not the full document) and returns findings that feed into the dimensional analysis.
Score each of the 8 dimensions on a 1-5 scale. Read references/review-checklist.md for the detailed rubric per dimension and score level.
For each dimension:
The 8 dimensions:
| Dimension | What it checks |
|---|---|
| Completeness | All critical sections present? Missing sections flagged. Full: 15. Summary: 9. |
| Clarity | Scope unambiguous? Requirements measurable? No vague language? |
| Consistency | Deliverables in scope match deliverables table? Milestones aligned with timeline? |
| Risk Coverage | Risks identified? Mitigations actionable? Assumptions explicit? Dependencies documented? |
| Commercial Adequacy | Payment terms clear? Change process defined? Penalties proportionate? |
| Collaboration Model | Roles defined? RACI complete? Governance clear? Escalation paths documented? |
| Best Practice Adherence | Follows PMI/PMBOK patterns? Exclusions explicit? Acceptance criteria per deliverable? |
| Corporate Standards | If provided: formatting, terminology, required clauses, branding compliance |
If no corporate standards are provided, score Corporate Standards as N/A and compute the overall grade from 7 dimensions.
Role-play two adversarial perspectives. Read references/common-pitfalls.md for known exploitation patterns.
As the Client:
As the Vendor:
For each exploitable gap, describe the scenario and recommend a specific fix.
Produce a structured review report with these sections:
| Dimension | Score (1-5) | Key Evidence |
|---|---|---|
| Completeness | {SCORE} | {EVIDENCE} |
| Clarity | {SCORE} | {EVIDENCE} |
| Consistency | {SCORE} | {EVIDENCE} |
| Risk Coverage | {SCORE} | {EVIDENCE} |
| Commercial Adequacy | {SCORE} | {EVIDENCE} |
| Collaboration Model | {SCORE} | {EVIDENCE} |
| Best Practice Adherence | {SCORE} | {EVIDENCE} |
| Corporate Standards | {SCORE} or N/A | {EVIDENCE} |
| Overall | {AVG} |
Overall grade interpretation:
Must-fix items before signing: missing sections, contradictions, legal gaps, unbounded obligations.
Should-fix items for quality: vague language, missing traceability, weak acceptance criteria, incomplete RACI.
"If I were the client/vendor, I would exploit..." scenarios with specific fixes.
Checklist of corporate standard requirements with pass/fail per item.
Save the report to docs/outbox/<project-name>-sow-review.md.
| Step | Documents to Read |
|---|---|
| Steps 1-2 | (no references — document ingestion and research) |
| Step 3 | references/review-checklist.md + references/best-practices.md |
| Step 4 | references/common-pitfalls.md |
| Step 5 | (no references — report generation) |
sow-write: run sow-review on the generated SOW to identify gaps before sharing with stakeholderssow-estimate: review ensures the SOW's scope and phases are well-defined enough for WBS extractionsow-write for targeted revisionsnpx claudepluginhub mrbogomips/claude-code --plugin project-managementCreates, edits, and optimizes skills for Claude Code, including drafting, evaluating with test prompts, iterating on performance, and improving skill descriptions for better triggering accuracy.