specalign
Align spec files with implementation. Detects drift between spec and code, surfaces discrepancies, user decides whether to update spec or code. Use when both a spec file and its implementation are in context.
90
Quality
87%
Does it follow best practices?
Impact
92%
1.12xAverage score across 3 eval scenarios
Passed
No known issues
Evaluation results
100%
14%Spec Alignment Review for Auth Module
Drift detection and DRIFT-NN reporting format
DRIFT-NN IDs used
0%
100%
Spec says quoted
100%
100%
Code does described
100%
100%
Impact stated
100%
100%
Mapping table present
100%
100%
Drift type categorized
50%
100%
User decision options presented
100%
100%
No files modified
100%
100%
Type drift detected
100%
100%
Behavior drift detected
100%
100%
Missing spec detected
100%
100%
Error handling drift detected
100%
100%
Without context: $0.1979 · 1m 10s · 8 turns · 13 in / 3,881 out tokens
With context: $0.2822 · 1m 23s · 11 turns · 447 in / 4,820 out tokens
99%
4%Apply Spec Alignment Decisions for Config Loader
Mutation policy and alignment summary
DRIFT-01 code fix
100%
100%
DRIFT-02 range fix
100%
100%
DRIFT-03 spec updated
100%
100%
DRIFT-04 skipped
100%
100%
Spec structure preserved
100%
100%
Lint/compile check run
87%
87%
Summary file present
100%
100%
Summary counts correct
90%
100%
No extra changes
100%
100%
Summary remaining section
100%
100%
Summary resolved counts
70%
100%
Without context: $0.3724 · 1m 23s · 21 turns · 28 in / 5,200 out tokens
With context: $0.4156 · 1m 13s · 22 turns · 25 in / 4,619 out tokens
79%
13%Add Low-Stock Alert Function to Inventory Module
Passive drift detection during unrelated task
Drift not ignored
100%
100%
DRIFT-NN format used
0%
100%
Spec says vs Code does
100%
100%
User decision presented
0%
0%
No unilateral fix of drift
0%
30%
New function in spec
100%
100%
New function implemented
100%
100%
Spec structure preserved
100%
100%
Implementation notes file
100%
100%
Drift impact stated
42%
100%
Passive framing
100%
33%
Without context: $0.3233 · 1m 11s · 21 turns · 28 in / 4,051 out tokens
With context: $0.3903 · 1m 23s · 22 turns · 304 in / 4,849 out tokens
- Repository
- NeverSight/skills_feed
- Commit
5342bca
- Evaluated
- Agent
- Claude Code
- Model
- Claude Sonnet 4.6
Table of Contents
Is this your skill?
If you maintain this skill, you can claim it as your own. Once claimed, you can manage eval scenarios, bundle related skills, attach documentation or rules, and ensure cross-agent compatibility.