affaan-m/continuous-agent-loopv1.2

continuous-agent-loop

Patterns for continuous autonomous agent loops with quality gates, evals, and recovery controls.

global

0installs0uses—~255

v1.2Saved Apr 20, 2026

Continuous Agent Loop

This is the v1.8+ canonical loop skill name. It supersedes autonomous-loops while keeping compatibility for one release.

Loop Selection Flow

Start
  |
  +-- Need strict CI/PR control? -- yes --> continuous-pr
  |
  +-- Need RFC decomposition? -- yes --> rfc-dag
  |
  +-- Need exploratory parallel generation? -- yes --> infinite
  |
  +-- default --> sequential

Combined Pattern

Recommended production stack:

RFC decomposition (ralphinho-rfc-pipeline)
quality gates (plankton-code-quality + /quality-gate)
eval loop (eval-harness)
session persistence (nanoclaw-repl)

Failure Modes

loop churn without measurable progress
repeated retries with same root cause
merge queue stalls
cost drift from unbounded escalation

Recovery

freeze loop
run /harness-audit
reduce scope to failing unit
replay with explicit acceptance criteria

Files1

1 files · 1.0 KB

Select a file to preview

Overall Score

42/100

Grade

Below Average

Safety

Quality

Clarity

Completeness

Summary

This skill provides architectural guidance for designing continuous autonomous agent loops with quality gates, evaluation harnesses, and failure recovery mechanisms. It documents a selection flow for choosing between different loop patterns (sequential, continuous-pr, rfc-dag, infinite) and recommends a production stack combining RFC decomposition, code quality gates, evaluation loops, and session persistence.

Detected Capabilities

Loop pattern selection and recommendationQuality gate integration guidanceFailure mode identificationRecovery procedure documentationProduction stack compositionRFC-based decomposition support

Trigger Keywords

Phrases that MCP clients use to match this skill to user intent.

autonomous agent loopquality gatesagent failure recoveryloop pattern selectioncontinuous evaluation

Use Cases

Design autonomous agent workflows with quality controls
Select appropriate loop pattern for CI/PR integration
Implement failure detection and recovery in agent systems
Set up evaluation harnesses and quality gates for agent outputs
Debug and audit continuous agent execution chains

Quality Notes

POSITIVE: Clear decision flow diagram aids pattern selection
POSITIVE: Specific recommended stack identifies concrete dependencies (ralphinho-rfc-pipeline, plankton-code-quality, eval-harness, nanoclaw-repl)
NEGATIVE: No implementation examples or code snippets — skill is purely architectural reference
NEGATIVE: Failure modes are listed but lack detailed troubleshooting steps or metrics for detecting each condition
NEGATIVE: Recovery procedures are terse; 'reduce scope to failing unit' lacks actionable detail on how to identify or isolate the unit
NEGATIVE: No configuration parameters, environment setup, or integration prerequisites documented
NEGATIVE: Cross-references to dependent skills (ralphinho-rfc-pipeline, eval-harness, etc.) lack URLs or integration instructions
NEGATIVE: Missing edge cases: how to handle cascading failures, partial state loss, or concurrent loop instances
NEGATIVE: No monitoring or observability guidance — no mention of logging, metrics, or dashboards for production loops
NEGATIVE: Version note ('v1.8+ canonical loop skill name') suggests breaking changes or migration burden, but no migration guide provided
NEGATIVE: Completeness issue: skill reads as architectural blueprint but lacks depth for practical implementation

Model: claude-haiku-4-5-20251001Analyzed: Apr 20, 2026

Reviews

Add this skill to your library to leave a review.

No reviews yet

Be the first to share your experience.

Version History

v1.2

Content updated

2026-04-20

Latest

v1.1

Content updated

2026-04-12

v1.0

Seeded from github.com/affaan-m/everything-claude-code

2026-03-16

Add affaan-m/continuous-agent-loop to your library and activate them in your IDE via MCP.

Get Started Free