0xrdan/retry
skills/retry/SKILL.md
Retry the last query with an escalated model
$ install --global
skillsauthnpx skillsauth add 0xrdan/claude-router retryInstall 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%
Error
VirusTotalMulti-engine malware detection
70%
Skipped
Snyk (dep)Open source security scanning
50%
Skipped
Socket.devSupply chain security analysis
50%
Skipped
CrowdStrikeAdvanced threat intelligence
50%
Skipped
OSV-ScannerOpen Source Vulnerability database check
50%
Skipped
OWASP Dep-Check
50%
Last scanned: Mar 24, 2026, 2:07 PM41.5s1 file scanned
SKILL.md
- name:
- retry
- description:
- Retry the last query with an escalated model
- user_invokable:
- true
Retry Skill
Retry the last query with a more capable model when the initial attempt failed or was insufficient.
What This Does
When a query routed to a cheaper model (Haiku or Sonnet) produces unsatisfactory results, errors, or timeouts, use /retry to:
- Re-route to the next tier up (Haiku -> Sonnet -> Opus)
- Preserve the original query context
- Give the more capable model a chance to succeed
Usage
/retry # Escalate to next tier
/retry deep # Force escalation to Opus
/retry standard # Force escalation to Sonnet
When to Use
- Timeout or error: Query failed due to complexity
- Incomplete answer: Model didn't fully address the question
- Wrong approach: Model misunderstood the task
- Need more depth: Initial answer was too superficial
How It Works
- Read the last routing decision from session state (
~/.claude/router-session.json) - Determine the appropriate escalation:
fast(Haiku) ->standard(Sonnet)standard(Sonnet) ->deep(Opus)deep(Opus) -> Already at max, suggest different approach
- Re-execute the last query with the escalated model
- Update session state with the new route
Instructions
When this skill is invoked:
- Check for explicit model - Did user specify
deep/opusorstandard/sonnet?- YES: Use that model. DO NOT auto-escalate. Honor the explicit choice.
- NO: Proceed to escalation logic
- Read session state from
~/.claude/router-session.json - Determine escalation (only if no explicit model):
- From
fast: Escalate tostandard - From
standard: Escalate todeep - From
deep: Already at max, suggest different approach
- From
- Inform the user of the escalation
- Spawn the appropriate subagent using Task tool
CRITICAL: If user specifies a model (/retry deep, /retry standard), use exactly that model. No exceptions.
Example
User ran a complex refactoring query that was routed to Haiku. Haiku produced incomplete results.
User: /retry
Assistant: Escalating from Haiku to Sonnet...
Re-running your refactoring query with more capable model.
[Spawns standard-executor with the original query]
Notes
- This skill reads the session state, which persists for 30 minutes
- If no previous query exists, inform the user
- Consider the failure reason when suggesting the escalation level
Related Skills
0xrdan/router-stats
tools
VerifiedTrustedCommunity
Display Claude Router usage statistics and cost savings
33SKILL.mdUpdated Mar 24, 2026
0xrdan/router-plugins
tools
VerifiedTrustedCommunity
List and toggle official plugin integrations
33SKILL.mdUpdated Mar 24, 2026
0xrdan/router-analytics
development
VerifiedTrustedCommunity
Generate HTML analytics dashboard for routing statistics
33SKILL.mdUpdated Mar 24, 2026
0xrdan/route
data-ai
VerifiedTrustedCommunity
Manually route a query to the optimal Claude model (Haiku/Sonnet/Opus)
33SKILL.mdUpdated Mar 24, 2026