tkersey/review-adjudication
codex/skills/review-adjudication/SKILL.md
Convert review claims into minimal, intent-anchored counterexamples. Verify current behavior, branch liability, AC-v2 horizon relation, novelty, kernel impact, and the only legal disposition. Use for review findings, PR comments, CAS findings, terminal holdouts, CEX-v1, or deciding whether a valid issue belongs in the current campaign. Never issue direct code-mutation authority or hand raw review prose to an implementer.
$ install --global
skillsauthnpx skillsauth add tkersey/dotfiles review-adjudicationInstall this skill globally with one command. Works with Claude Code, Cursor, and Windsurf.
Security Scan Results
3 of 9 scanners reported clean
Some scanners were skipped, did not run, or reported a non-clean status. Review each row below.
3
Scanners Passed
9
Scanners in report
Clean
TrivyContainer and dependency vulnerability scanner
95%
Clean
SemgrepStatic code analysis for vulnerabilities
95%
Clean
mcp-scan (Snyk)Model Context Protocol security validation
95%
Skipped
Snyk (dep)Open source security scanning
50%
Skipped
Socket.devSupply chain security analysis
50%
Skipped
VirusTotalMulti-engine malware detection
50%
Skipped
CrowdStrikeAdvanced threat intelligence
50%
Skipped
OSV-ScannerOpen Source Vulnerability database check
50%
Skipped
OWASP Dep-Check
50%
Last scanned: Jun 29, 2026, 4:34 AM328.8s57 files scanned
SKILL.md
- name:
- review-adjudication
- description:
- Convert review claims into minimal, intent-anchored counterexamples. Verify current behavior, branch liability, AC-v2 horizon relation, novelty, kernel impact, and the only legal disposition. Use for review findings, PR comments, CAS findings, terminal holdouts, CEX-v1, or deciding whether a valid issue belongs in the current campaign. Never issue direct code-mutation authority or hand raw review prose to an implementer.
- version:
- 5.0.0
- activation_cost:
- medium
- default_depth:
- strict
Review Adjudication
Mission
Separate:
observed fact
review claim
repair suggestion
branch liability
intent entailment
counterexample novelty
kernel impact
legal disposition
Output one counterexample / CEX-v1 or a rejected/blocked record.
A valid diff-local issue is not automatically campaign scope.
Required inputs
current artifact state
review mode
AC-v2 ID/fingerprint/horizon
accepted kernel/law refs when present
review claim and source refs
current code/proof evidence
known counterexample classes
If AC-v2 is required but absent or stale:
disposition = blocked
Review modes
discovery
kernel_review
conformance
terminal_holdout
Discovery
Broad search is allowed. Findings still require AC relation before admission.
Kernel review
Evaluate AC/CEX/CEB/MBK/RC consistency. Do not review delivery implementation.
Conformance
Only named RAP-v1 apertures are in review scope.
Allowed outputs:
existing-law violation
missing law proof
orphan construct
stale artifact state
novel in-horizon CEX
outside-horizon proposal
clean
Terminal holdout
Broad adversarial search is allowed once. No direct mutation authority.
CEX-v1
counterexample:
counterexample_version: CEX-v1
counterexample_id:
campaign_id:
batch_id:
aperture_id:
review_mode:
artifact_state:
base:
head:
dirty_fingerprint:
review_receipt:
claim:
statement:
source_refs: []
suggested_repair:
observation:
actor:
operation:
pre_state:
minimal_trace: []
expected:
actual:
externally_visible_difference:
reproduction_or_proof:
validity:
confirmed |
refuted |
stale |
unknown
liability:
introduced_by_current_diff |
exposed_and_required_by_current_acceptance |
preexisting_but_blocks_current_invariant |
adjacent_preexisting |
reviewer_preference |
unknown
intent:
contract_id:
contract_fingerprint:
horizon_fingerprint:
acceptance_refs: []
compatibility_refs: []
forbidden_refs: []
non_goal_refs: []
kernel_law_refs: []
relation:
directly_entailed |
compatibility_required |
forbidden_state_witness |
contract_invalidating |
outside_horizon |
unrelated |
unknown
witness:
scope_effect:
none |
narrows |
expands |
invalidates_contract
novelty:
new_equivalence_class |
new_witness_existing_class |
duplicate |
refuted |
stale |
unknown
existing_class_ref:
kernel_impact:
existing_law_violation |
additional_witness |
missing_semantic_distinction |
missing_proof |
orphan_realization |
stale_artifact_state |
no_kernel_impact |
unknown
disposition:
enter_kernel |
attach_witness |
invalidate_realization |
return_to_contract |
capture_followup |
reject |
blocked
mutation_authority:
allowed: no
reason:
Mutation authority is always no at the finding level.
The controller grants realization authority only after a batch is sealed and the kernel/design gates pass.
Minimal distinguishing trace
A confirmed actionable CEX must identify the smallest trace that distinguishes expected from actual behavior:
actor
operation
pre-state
transition sequence
observable result
A broad code smell, speculative risk, or preferred repair is not a minimal trace.
Deterministic decision table
Refuted or stale
disposition = reject
Unknown validity, liability, or intent relation
disposition = blocked
Reviewer preference or unrelated claim
disposition = reject
Adjacent preexisting + outside horizon
disposition = capture_followup
Contract invalidating or scope expanding
disposition = return_to_contract
Missing semantic distinction inside sealed horizon
novelty = new_equivalence_class
disposition = enter_kernel
Existing-law violation during conformance
disposition = invalidate_realization
No direct patch authorization.
New witness for existing class
disposition = attach_witness
No new code distinction or test family.
Duplicate
disposition = reject
or attach_witness only when it materially strengthens proof
Missing proof
disposition = attach_witness
Proof-only by default.
Orphan realization
disposition = invalidate_realization
Novelty discipline
new_equivalence_class requires a witness that an existing class cannot explain.
Similarity to a different file, function, or review comment is not novelty.
new_witness_existing_class cannot authorize:
new branch
new helper
new state
new protocol case
new fallback
new public symbol
new wound-specific test family
Current-code verification
Before classification:
- Reproduce or prove the observed behavior.
- Identify the authority owner.
- Distinguish claim from suggested repair.
- Compare against current AC/kernel fingerprints.
- Search existing CEX classes.
- Classify liability and horizon relation.
- Derive disposition.
Do not accept stale review text as current evidence.
Validation
python3 codex/skills/review-adjudication/tools/counterexample_gate.py cex.json
Output
Return:
counterexample ID
validity / liability
intent relation / anchors
minimal trace
novelty / existing class
kernel impact
disposition
mutation authority = no
source/proof refs
Hard rules
- Valid does not mean branch-liable.
- Branch-liable does not mean intent-entailed.
- Repair suggestions are not scope.
- Raw review prose never reaches an implementer.
- Every admitted CEX has a minimal trace and stable intent/law anchors.
- Additional witness does not create a code distinction.
- Conformance findings invalidate or return; they do not directly patch.
- Scope expansion returns to AC authority.
- Every finding-level mutation authority is
no.
Related Skills
tkersey/fm
tools
VerifiedTrustedCommunity
Invokes Apple's macOS 27 fm command-line tool from a local Mac to use the on-device system model or Private Cloud Compute, including instructions, image prompts, schema-constrained JSON, and noninteractive automation. Use when the user asks to run Apple Foundation Models through fm, compare system versus pcc, generate structured output, or automate fm without Swift or an app.
64SKILL.mdUpdated Jul 20, 2026
tkersey/hylo
development
VerifiedTrustedCommunity
Compile historical Codex sessions into governed counterfactual evidence, evaluate an existing owner-applied candidate through blinded paired HCTP trials, and fold observable evidence into RUN, OBSERVE, or STOP. Use for `$hylo`, CRF extraction, counterfactual replay, source-governed direct or historical trials, sealed evidence, paired baseline/candidate evaluation, causal frontiers, or evidence-governed improvement.
64SKILL.mdUpdated Jul 13, 2026
tkersey/ledger
testing
VerifiedTrustedCommunity
Ensure a `ledger` command is available on PATH; materialize, validate, record, replay, and project requested Actuating artifacts without taking semantic or execution authority; coordinate the shared Learnings/Synesthesia/Negative Ledger lifecycle checkpoint and repo-local source-memory reconciliation; address Universalist plans and receipts; and perform pure artifact validation.
64SKILL.mdUpdated Jun 29, 2026
tkersey/review-fold
testing
VerifiedTrustedCommunity
Classify and quotient review findings, failing tests, incidents, bug reports, migration failures, and other witnessed falsifiers against accepted intent and the current Construction. Author counterexample-set/v1 without selecting repairs, counting review credit, or granting mutation.
64SKILL.mdUpdated Jun 28, 2026