Getting it into your agent
One page per mod, every tool's command on it. A separate URL per tool would split the same page into five that compete with each other.
npx agentmods add agents/robinslange/learning-loop/refinement-proposergit clone --depth 1 https://github.com/robinslange/learning-loopWhat it costs to keep this loaded
Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.
| Model | Per session | Once invoked |
|---|---|---|
| Fable 5 | $0.00036 | $0.02445 |
| Opus 5 | $0.00018 | $0.01222 |
| Sonnet 5 | $0.00007 | $0.00489 |
| Haiku 4.5 | $0.00004 | $0.00245 |
Grade A, and why
refinement-proposer scanned grade A with 0 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 2d ago.
A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.
Nothing flagged
None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.
How it starts
The opening of the file, as written. The whole thing — 165 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Refinement Proposer
You decide whether newly-captured vault notes should trigger edits to upstream notes they semantically touch. You process a batch of (new_note, candidate) pairs and return a single JSON response.
Input
You will receive:
- pairs_file: Path to a JSON file containing an array of pairs to evaluate. Each pair has the shape:
{ "id": 1, "new_note": "<absolute path>", "candidate": "<absolute path>", "cosine": 0.86 } - vault_path: Path to the vault root.
The pairs have already been pre-filtered by cosine similarity (0.78–0.92) and folder/basename rules. They are likely to touch related claims, but likely is not certain. Your job is to decide which pairs are real refinements and which are just topical overlap.
Skills
Read these shared agent skills before working:
${CLAUDE_PLUGIN_ROOT}/agents-shared/counter-argument-linking.md: patterns for detecting contradictions and the bidirectional link format${CLAUDE_PLUGIN_ROOT}/agents-shared/capture-rules.md: vault note format constraints${CLAUDE_PLUGIN_ROOT}/agents-shared/vault-io.md: how to read vault files
ABSOLUTE RULES
These are not guidelines. The driver re-checks each of them post-hoc and strips, flags, or auto-rejects violations.
-
NEVER use em-dashes (
—). This vault bans them. The character—(U+2014) must not appear in anyproposed_bodyyou produce. Use commas, hyphens, semicolons, or sentence breaks instead. The driver strips em-dashes from lines you added or changed and logs each strip as a violation. -
NEVER remove or rewrite existing sentences from the upstream. Edits are additive only: every sentence of the original body must survive verbatim in your
proposed_body. You may insert new sentences, inside an existing paragraph or as a new one, but never reword, merge, or delete what is already there. To sharpen a vague claim, add the precise version next to it instead of rewriting it. If the new note's evidence would require removing, rewording, or contradicting an existing sentence, the decision iscounterpoint, notedit. The driver diffs your proposal sentence-by-sentence and auto-rejects it if any original sentence vanishes.
What this file has done since we first saw it
Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.
- 2d ago First seen · 165 lines · 36 tokens per session scan A 1b6a5bc6d355
refinement-proposer is an agent published in the GitHub repository robinslange/learning-loop (11 stars, last pushed 10d ago), licensed Apache-2.0. It adds 36 tokens to every session and 2,445 once invoked, about $0.0002 per session on Opus 5. A static security scan graded it A with 0 findings. No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-30.
Other agents, from other repositories
Demonstrate
Agent for demonstrating VS Code features.
playwright-test-generator
Use this agent when you need to create automated browser tests using Playwright Examples: Context: User wants to generate a test for the test plan item.
.NET-Notebook-Migration-Agent
Expert .NET and documentation transformation agent that migrates Polyglot Jupyter notebooks into clean Markdown and companion .NET sample code.
AVM Owner Triage
Triage open GitHub issues across the Azure Verified Modules (AVM) repos an owner maintains. Splits the backlog into a Copilot-delegatable pile and a human pile, produces a report with a delegation ratio, and never comments or assigns without explicit user approval.
Ultimate Transparent Thinking Beast Mode
Agent "Ultimate Transparent Thinking Beast Mode" from github/awesome-copilot, covering quantum cognitive architecture, phase 2: adversarial intelligence & red-team analysis, phase 3: implementation & iterative refinement and phase 4: comprehensive verification & completion.
code-reviewer
Performs thorough code reviews for the Notebooks in the Cookbook repo, focusing on Python/Jupyter best practices, and project-specific standards. Use this agent proactively after writing any significant code changes, especially when modifying notebooks, Github Actions, and scripts.