unsloth-mcp-server: Agent for Claude Code

.claude/agents/observer-full.md

observer-full is an agent for Claude Code from ScientiaCapital/unsloth-mcp-server. It costs 39 tokens per session (1,123 once invoked), scanned A, original, Apache-2.0.

A thorough review agent for the unsloth-mcp-server project that checks code quality and system design. It looks for seven kinds of project drift and records findings in OBSERVERQUALITY.md and OBSERVERARCH.md.

In plain words
What is it for?
Use it to review STANDARD or FULL-scope tasks for scope violations, TODO-style debt, missing tests, and architecture problems. It compares changes with feature contracts when those contracts exist.
Why use it?
It helps catch work that changes files outside the agreed scope, adds technical debt, or lacks tests. Its devil's-advocate approach is intended to question whether an implementation is complete.

Agent for Claude Code

Written for Claude Code: installed under .claude/. Also seen: model in frontmatter.

This is ScientiaCapital/unsloth-mcp-server's own configuration. It tells Claude Code how to work on unsloth-mcp-server itself, so it is not a mod to install elsewhere. Copy it as a starting point and replace the rules that are about this project. Everything unsloth-mcp-server configures →

Reuse

Borrowing it

Nothing to install: this file belongs to ScientiaCapital/unsloth-mcp-server. Take a copy, put it at the same path in your own repository, and replace the rules that are about this project with yours.

Copy the file
curl -O https://raw.githubusercontent.com/ScientiaCapital/unsloth-mcp-server/main/.claude/agents/observer-full.md
Clone the repo
git clone --depth 1 https://github.com/ScientiaCapital/unsloth-mcp-server

Made for: Claude Code.

Wrote this? Show the measurements

A badge with what this costs and how it scanned, read live from this page, so it follows the numbers instead of freezing them. Markdown for a README, HTML for a documentation site or a project page.

agentmods badge for observer-full

README.md
[![agentmods](https://agentmods.dev/badge/agents/scientiacapital/unsloth-mcp-server/observer-full.svg)](https://agentmods.dev/agents/scientiacapital/unsloth-mcp-server/observer-full)
Your own site
<a href="https://agentmods.dev/agents/scientiacapital/unsloth-mcp-server/observer-full"><img src="https://agentmods.dev/badge/agents/scientiacapital/unsloth-mcp-server/observer-full.svg" alt="Measured on agentmods" height="20"></a>
Per session 39 Only the description is in the session, so the agent can decide to use it. The body loads when it is invoked.
When invoked 1,123 The whole file, excluding the scripts and references it only reads on demand.
Security scan A 0 findings. A grade says what 26 rules found in the file — not that it is safe.
Origin original No closer match found in the catalogue.
Token cost

What 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.

ModelPer sessionOnce invoked
Fable 5.1 $0.00039 $0.01123
Opus 5 $0.00019 $0.00562
Sonnet 5 $0.00008 $0.00225
Haiku 4.5 $0.00004 $0.00112

Measured 6d ago against content hash c588bb46d73f, method: parsed. Prices are Anthropic first-party input rates as of 2026-09-06, from the pricing page.

Security

Grade A, and why

observer-full 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 6d 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.

.claude/agents/observer-full.md · 156 lines

How it starts

The opening of the file, as written. The whole thing — 156 lines — stays where its author put it; the contents beside it link to each section on GitHub.

Observer Full: Comprehensive Quality + Architecture Analysis

You are the full observer for the unsloth-mcp-server project, combining both Code Quality and Architecture Observer roles. You run all 7 drift detection patterns and maintain a devil's advocate stance.

Your Mandate

Assume builders WILL cut corners under pressure. Your job: prove them wrong.

7 Drift Detection Patterns

1. Agent Drift (Scope Violation)

Compare modified files against the feature contract (if one exists in .claude/contracts/):

git diff --name-only main...HEAD
  • Files modified outside the contract scope = [BLOCKER]
  • If no contract exists, note this as [WARNING]: "No feature contract defined"

2. Tech Debt Accumulation

Count TODO/FIXME/HACK/XXX/TEMP markers in changed files:

  • Use Grep across all source files
  • If count increases by >3 compared to what's reasonable for project size = [WARNING]
  • List each marker with file:line

3. Test Gaps

For every new/modified function or class, verify a corresponding test exists:

  • New exported functions without tests = [WARNING] for utilities
  • New API endpoints without tests = [BLOCKER]
  • Check both unit and integration test directories

4. Scope Creep

Detect features being built that aren't in the contract:

  • New API endpoints not in contract = [BLOCKER]
  • New files in unexpected directories = [WARNING]
  • Significant new functionality beyond stated scope = [WARNING]

5. Import Bloat

Detect unused imports and redundant dependencies:

  • Unused imports in TypeScript/JavaScript = [INFO]
  • New dependencies in package.json/pyproject.toml without justification = [WARNING]

6. Silent Failures

Detect empty catch blocks and swallowed errors:

  • Empty catch blocks = [WARNING]
  • Bare except in Python = [WARNING]
  • console.log in catch blocks without re-throwing = [INFO]

7. Contract Drift

If a feature contract exists, verify implementation matches:

Read the full file on GitHub · 156 lines

Changes

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.

  1. 6d ago First seen · 156 lines · 39 tokens per session scan A c588bb46d73f

Subscribe to this mod's changes

observer-full is an agent published in the GitHub repository ScientiaCapital/unsloth-mcp-server (2 stars, last pushed 5mo ago), licensed Apache-2.0. It adds 39 tokens to every session and 1,123 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 2026-08-31.

Related

Other agents, from other repositories

cpp-reviewer

Expert C++ code reviewer specializing in memory safety, modern C++ idioms, concurrency, and performance. Use for all C++ code changes. MUST BE USED for C++ projects.

affaan-m/ECC · 41 tokens

reviewer

Read-only reviewer for an SDD implementation — checks that the change satisfies the acceptance criteria it claims (stage 1) and meets quality/convention/edge-case bars (stage 2). Use after a task (or the whole feature) reaches GREEN, before it's considered done. It reads the diff and the upstream artifacts and reports…

genkovich/sdd · 81 tokens

atomic-auditor

Final gate for a finished implementation. Dispatched exactly once after the implement-review loop goes green, never per iteration. Never touches the repo; its one write is the audit report into the task scratchpad. Audits the delivered work as a whole: cumulative spec compliance, cross-iteration coherence…

damusix/atomic-claude · 169 tokens

bt6-pr-auditor

Reviews one pull request in a BT6 codebase for correctness, research integrity, security, verification quality, and merge readiness.

elder-plinius/T3MP3ST · 32 tokens

Reviewer

Mandatory fast reviewer: validates every agent delegation output before acceptance. Checks acceptance criteria, file partitions, regressions, type safety, security basics.

monkilabs/opencastle · 30 tokens

security-auditor

Use this agent when reviewing local code changes or pull requests to identify security vulnerabilities and risks. This agent should be invoked proactively after completing security-sensitive changes or before merging any PR.

NeoLabHQ/context-engineering-kit · 40 tokens