<!-- xid: 02E852427ED9 --> <a id="xid-02E852427ED9"></a>
Scanned 9/12/2026
Install to Claude Code
npx -y skills add synthaicode/XRefKit --skill review_report_composition --agent claude-codeInstalls into .claude/skills of the current project.
Are you the author of Review Report Composition?
Add the live security badge to your README — it updates automatically with every re-scan.
[](https://www.skillsdirectory.com/skills/synthaicode-review-report-composition)More formats (shields.io, HTML) on the badges page.
<!-- xid: 02E852427ED9 -->
<a id="xid-02E852427ED9"></a>
# Skill: review_report_composition
## Purpose
Compose review Skill outputs into a report that a human reviewer can use for
decision making.
This Skill does not detect technical defects. The detector Skill owns category
applicability, findings, severity, evidence, and remediation judgment. This
Skill owns whether those detector results are expressed clearly, without hiding
the judgment basis or changing the detector's conclusion.
## Inputs
- detector Skill name and reviewed scope
- category results and detector status for each active category
- findings with severity and evidence references
- usage premise or review premise when supplied
- optional draft report
## Outputs
- composed review report
- category matrix expression check
- finding expression check
- required-input result table when the detector produced required-input facts
- composition issues that require detector clarification
- handoff list for out-of-scope or under-evidenced report content
## Boundary
- Do not invent findings, categories, evidence, or remediations.
- Do not change detector severity or technical judgment unless the detector
output already contains evidence for that change.
- Do not use the user's headline purpose to suppress category rows.
- Do not treat report wording as proof that a category was reviewed.
- Do not place implementation-specific examples in this Skill. Target-specific
details belong in the composed report or eval fixtures, not in the canonical
instruction.
## Category Matrix Composition
Every active detector category must remain visible in the report.
Each category row must preserve these report slots when applicable:
- category
- status
- reviewed evidence
- judgment basis
- remaining unknown or validation boundary
- handoff target when the category cannot be closed by the detector
Assigning category status is the detector Skill's responsibility. This Skill
checks that the report expresses the detector's status in a decision-readable
way:
- `pass`: the report names the reviewed evidence and why no violation remains.
- `fail`: the report names the violated condition and affected behavior.
- `needs_confirmation`: the report names the missing evidence and decision that
cannot be closed.
- `not_applicable`: the report names the absence basis for the category's own
review axis.
`not_applicable` is not a valid expression merely because the category is
unrelated to the headline review purpose, unrelated to the final root cause, or
not represented by a finding.
## Finding Composition
Each finding must preserve the detector's evidence and make the review decision
auditable. Use these slots as the expression contract:
- finding id
- detector source
- category
- severity
- observed condition
- required precondition, expected invariant, or violated rule
- affected state, decision, contract, or execution path
- consequence or review risk
- evidence reference
- recommended remediation or next owner
- unresolved unknowns
The exact wording can vary by report format. The slots must not be collapsed
into summary-only language when the collapsed form prevents a reviewer from
understanding why the issue is a defect, why the severity was chosen, or what
must be changed.
If the detector output lacks a slot needed to support the conclusion, do not
guess. Record a composition issue and hand the item back to the detector Skill
for clarification.
## Required Input Result Composition
When a detector reports required business input integrity, express the result
as rows that make the decision basis visible:
| input / candidate | decision gated | source | missing or invalid behavior | default provenance | disposition | status |
|---|---|---|---|---|---|---|
- `input / candidate`: the reviewed input, or an explicit absence row when no
candidate exists in the reviewed scope.
- `decision gated`: the decision controlled by that input, or none when the
reviewed scope has no such decision.
- `source`: the source class used by the detector, or none.
- `missing or invalid behavior`: the detector's observed failure, fallback, or
controlled behavior.
- `default provenance`: whether the detector found the value to be configured,
derived, invented, absent, or unknown.
- `disposition`: the detector's decision for that row.
- `status`: the category status contribution for that row.
Do not express this category as only "business input is present/absent",
"library scope", or another summary phrase that hides the decision basis.
## Handoff Composition
Handoff rows must separate:
- unresolved detector evidence
- out-of-scope review category
- required runtime, integration, or manual validation
- ownership transfer to another Skill or workflow
Each handoff row must name the reason, target owner or Skill when known,
evidence reference, and what decision remains blocked.
## Monitoring And Control
- Check that every detector category is present in the report.
- Check that every finding has evidence and a reviewable decision basis.
- Check that category status wording does not contradict detector status.
- Check that summary phrasing has not replaced required judgment slots.
- Check that `needs_confirmation` and handoffs identify the missing evidence.
- Check that report composition issues are not silently rewritten as technical
findings.
## Closure
Closure is allowed when the composed report preserves detector judgments,
exposes decision bases for active categories and findings, and lists all
composition issues or detector handoffs that remain unresolved.
## Reporting Contract (共通報告)
- reporting_profile: artifact_traceability
Use the shared [Skill Reporting Contract](../../docs/core/contracts/081_skill_reporting_contract.md#xid-6B2D9F4A1C73) in the final report. Start with these headings in this order:
1. Status — done, partial, blocked, or escalated
2. Result — what was produced or decided
3. Evidence — output, evidence, checks, or XIDs
4. Open Items — unresolved unknowns, risks, judgments, or なし
5. Handoff — next owner and next action, or なし
Keep this summary-first section visible before Skill-specific detail; do not omit empty sections.
Is this your skill, or is something wrong with this listing? Request removal or report an issue. Author removals are honored within 72 hours.
No comments yet. Be the first to comment!