Pattern for progressively refining context retrieval to solve the subagent context problem
48
35%
Does it follow best practices?
Impact
Pending
No eval scenarios have been run
Passed
No known issues
Optimize this skill with Tessl
npx tessl skill review --optimize ./docs/zh-TW/skills/iterative-retrieval/SKILL.mdIterative retrieval cycle
Multi-phase cycle
70%
100%
Broad initial query
40%
100%
Numeric relevance scale
50%
100%
Relevance score bands
0%
100%
Exclude low-relevance files
0%
100%
Maximum 3 cycles
0%
100%
Return threshold ≥ 0.7
0%
100%
Early termination condition
30%
100%
Missing context tracking
0%
100%
Structured query object
0%
50%
Keyword expansion in refinement
100%
100%
Agent retrieval instructions
Broad initial search
87%
100%
Numeric relevance scoring
40%
100%
Include threshold ≥ 0.7
0%
100%
Maximum cycle count
0%
100%
Early stopping condition
90%
100%
Terminology learning
50%
100%
Gap identification
90%
100%
Search term expansion
80%
100%
Exclude irrelevant files
80%
100%
Minimum high-relevance count
28%
100%
Patterns and keywords together
80%
100%
Query refinement logic
Extract patterns from high-relevance
80%
100%
Extract keywords from high-relevance
80%
100%
Merge patterns cumulatively
100%
100%
Merge keywords cumulatively
100%
100%
Build focusAreas from gaps
0%
73%
Exclude low-relevance paths
53%
100%
All four query fields present
0%
70%
focusAreas deduplication
0%
100%
Exclusion threshold < 0.2
0%
100%
ae2cadd
Table of Contents
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.