Ready to supercharge your AI agents?

Install any skill in seconds. Free to start, no credit card required.

Get Started Free →

Self Refine Loop · stijnman | DecimalAI

agent toolingby stijnman

Self Refine Loop

stijnman/self-refine-loop

Source stijnman/grok-custom-skills ↗License MIT

Runs a generator-critique-reviser loop to iteratively improve agent outputs.

SkillScore73Maintained

Automatic

AI Engineer pack

Last verified 8/15/2026

What it changes

Passed 16 of 22 test cases with the skill, and 9 of 22 without it.

Without skill41%

With skill73%

Checks it passed that the model alone missed

Halts iteration early when confidence remains below 5
Contains no more than 5 bullet points in the critique section
Identifies the impossible constraints as a blocker when halting

and 5 more in the full report

7 fixed9 passed either way6 still failing0 broke

Pass-rate lift: +32 pp
Tokens: -7%
Agent turns: 0%
Response time: -21%

Model gemini-3.6-flashMeasured 8/15/2026v216 of 22 cases passed with skill

Tokens, turns and time describe resource use, not a dollar price.

Use this skill

Save it into your project's skills folder. Claude Code picks it up in its next session.

curl -fsSL --create-dirs -o .claude/skills/stijnman-self-refine-loop/SKILL.md https://app.decimal.ai/s/stijnman-self-refine-loop/SKILL.md

Want it pinned and updatable? pip install "decimalai>=0.15", then decimalai skills add stijnman-self-refine-loop, and start Claude Code with claude --plugin-dir "$(git rev-parse --show-toplevel)/.agents"

1 · One-time — install the CLI (docs ↗)

pip install "decimalai>=0.15"

2 · Get the skill — writes .agents/skills/stijnman-self-refine-loop/ and pins it in .decimal/skills.lock

decimalai skills add stijnman-self-refine-loop

No CLI? Get the same file — curl · raw file

curl -fsSL --create-dirs -o .claude/skills/stijnman-self-refine-loop/SKILL.md https://app.decimal.ai/s/stijnman-self-refine-loop/SKILL.md

Raw file: /s/stijnman-self-refine-loop/SKILL.md ↗ · machine-readable summary at /s/stijnman-self-refine-loop.json ↗

Safety checks: 3 of 3 passed

Code scan CleanIntent PassedContent Clean

See it in action

One of 7 test cases the skill fixed. Same model, same prompt, with and without the skill.

1 of 7

Prompt

Fix the spelling error in this sentence: 'Our database querys are optimized for speed.'

What was checked, without and with the skill
Checked	Without skill	With skill
Outputs 'queries' directly without displaying critique bullet points or numeric confidence scores	Missed	Passed

case-12 · gemini-3.6-flash · 8/15/2026 · answers are capped and redacted

What's inside

The instructions your model reads when the skill loads.

SKILL.md38 linessingle file

Self Refine Loop

When to Use

User says self refine or task matches this capability
User says reflexion loop or task matches this capability
User says critique and revise or task matches this capability
User says improve output or task matches this capability

Workflow

Capture the current output and the user's quality criteria.
Generate a critique listing specific weaknesses (max 5 bullets).
Revise the output addressing every critique point.
Score confidence 0-10 on whether criteria are met.
Repeat until confidence >= 8 or 5 iterations; return best version with changelog.

Integrations

goal-verifier
agentic-uncertainty-quantifier
dspy-prompt-optimizer

Error Handling

Failure	Response
No criteria given	Ask user for 1-3 success criteria before looping.
Confidence stuck below 5	Stop early; report blocker and ask for guidance.
Output grows unbounded	Cap revisions to prior length + 20%.

Gotchas

Do not loop on trivial typos; one-pass fix is enough.

Example

Input: User request matching triggers above. Output: Structured result per workflow with integrations invoked as needed.

Track record

Benchmark

16 / 22

test cases passed, vs 9 / 22 without

Adoption

Users: 0
Times used: 0

Safety and source details

SkillSafety

✓ 3 of 3 passed

Code scan15 of 15 checks clean

0 critical · 0 warning · 0 info — scanner v5 — 8/15/2026

View 15 scanner checks

live_secretremote_code_executionreverse_shelldata_exfiltrationssrf_metadataagent_snoopinganti_refusaldestructive_commandsobfuscated_unicodeunicode_homoglyphopaque_blobsuspicious_urltool_overreachtrigger_abuseprompt_injection_phrasing

IntentSafe

LLM judge · does the skill do what it claims

ContentClean

LLM judge · policy & harmful-content review

Source

Imported

stijnman / grok-custom-skills

.grok/skills/self-refine-loop/SKILL.md

View on GitHub ↗

License MITImported 8/15/2026Version v2

More skills from the same source and nearby in the registry, with their headline benchmark lift.

Ready to supercharge your AI agents?

See it in action

What's inside

Self Refine Loop

When to Use

Workflow

Integrations

Error Handling

Gotchas

Example

Track record

Benchmark

Adoption

SkillSafety

Source

Related skills