What is claude-gladiator?

A Gladiator plugin that observes tool failures and prompts reflection to evolve rules, hooks, and artifacts, with synergy from Historian and Oracle.

How does it observe and reflect?

Tool failures trigger automatic observations via PostToolUse; Stop prompts reflection if there are unprocessed observations. gladiator_reflect() clusters into recommendations, checks existing artifacts to avoid duplicates, and surfaces updates with reasoning. Historian and Oracle enrich reflections when active, and /review-gladiator can batch across sessions.

How do I install and start using it?

Run the command claude mcp add gladiator -- npx claude-gladiator-mcp to install, then enable the hooks (PostToolUse, Stop). Use /review-gladiator [topic] or gladiator_reflect() to process and apply updates after approval.

claude-gladiator

Scanned

npx machina-cli add skill Vvkmnn/claude-emporium/claude-gladiator --openclaw

Files (1)

SKILL.md

3.3 KB

Gladiator Plugin

Continuous learning. Observes tool failures and prompts reflection at session end to evolve rules, hooks, and skills.

Hooks

Hook	When	Action
PostToolUse(Bash\|Edit\|Write)	After tool failure	Observes the error pattern (silent on success)
Stop	Session ending	Prompts reflection if unprocessed observations exist

Commands

Command	Description
`/review-gladiator [topic]`	Batch learn from accumulated observations and session history

Workflows

Observe (automatic via hooks)

Tool failures trigger observation automatically:

gladiator_observe(
  summary: "<what failed and how it was fixed>",
  context: {error, tool, before, after},
  tags: ["error", "<tool_name>"]
)

Reflect (standalone)

gladiator_reflect() — cluster observations into recommendations
For each recommendation: read the existing artifact (if overlap detected)
Propose UPDATE to existing artifact, not a new duplicate
Present to user with reasoning
Apply changes one at a time after approval

Reflect (with siblings)

If historian active: enrich reflection with broader context
- search_conversations("project or topic") — related past work
- get_error_solutions("specific error") — for error clusters
- find_tool_patterns("tool name") — for tool workflow clusters
gladiator_reflect() — cluster observations
If oracle active: for each recommendation involving new artifacts
- search("cluster tag") — check if best-in-class solution already exists
- Install existing solution instead of reinventing
Present enriched recommendations: pattern + history + available tools
Apply changes one at a time after approval

Batch Review (/review-gladiator)

If historian active: list_recent_sessions() to get session refs
gladiator_observe(source: "conversation", session_ref: <ref>) for relevant sessions
gladiator_reflect() to cluster all observations
If oracle active: search for existing solutions before creating new
Present recommendations to user

Sibling Synergy

Sibling	Value	How
Historian	Past solutions enrich reflection	`get_error_solutions()`, `search_conversations()`, `find_tool_patterns()`
Oracle	Existing tools found before creating new	Search oracle for best-in-class solutions during reflection
Praetorian	n/a	Gladiator has its own persistence
Vigil	n/a	Different concerns (files vs patterns)

Observation Templates

Situation	Call
Tool failure (auto)	`gladiator_observe(summary, context={error, tool, before, after}, tags=["error", tool])`
User correction	`gladiator_observe(summary, context={before, after}, tags=["correction"])`
Convention found	`gladiator_observe(summary, tags=["convention", "domain"])`
Decision made	`gladiator_observe(summary, tags=["architecture", "decision"])`

Requires

claude mcp add gladiator -- npx claude-gladiator-mcp

Source

git clone https://github.com/Vvkmnn/claude-emporium/blob/main/plugins/claude-gladiator/skills/claude-gladiator/SKILL.mdView on GitHub

Overview

Claude-gladiator is a Gladiator Plugin that continuously learns from tool failures by observing outcomes after each tool use and prompting reflection at session end. It coordinates with sibling roles like Historian and Oracle to enrich analyses and prevent duplicate artifacts.

How This Skill Works

Tool failures trigger automatic observation through PostToolUse; the Stop hook prompts reflection if unprocessed observations remain. gladiator_reflect() clusters observations into recommendations, compares them to existing artifacts, and surfaces updates with reasoning for approval. With Historian and Oracle active, reflections are enriched with past conversations and best-in-class solutions, and /review-gladiator batches across sessions.

When to Use It

After a tool failure triggers an error that should be learned from
At session end when there are unprocessed observations to reflect on
When you need to avoid creating duplicate artifacts by checking existing ones first
When Historian or Oracle are active to enrich reflections with broader context and best practices
When batching learning across multiple sessions using /review-gladiator

Quick Start

Step 1: Install and activate the Gladiator plugin (requires: claude mcp add gladiator -- npx claude-gladiator-mcp)
Step 2: Ensure PostToolUse and Stop hooks are live so tool failures are automatically observed
Step 3: After a session, run gladiator_reflect() or /review-gladiator [topic] and approve updates one at a time

Best Practices

Capture a clear observation with summary and context: error, before, after
Before proposing updates, read the existing artifact to avoid duplicates
Run gladiator_reflect() to cluster observations before applying changes
Leverage Historian/Oracle to enrich reflections with past work and best solutions
Apply changes one at a time after user approval; use /review-gladiator for batch reviews

Example Use Cases

After a Bash/Edit failure, gladiator_observe stores the error and later suggests updating a Bash error-handling artifact
On session end, Stop triggers reflection to refine rules based on accumulated observations
With Historian active, get_error_solutions and search_conversations enrich the reflection for a failing network call
With Oracle active, search for best-in-class solutions before creating a new artifact and install it if found
Use /review-gladiator to batch-review observations across recent sessions and unify solutions

Frequently Asked Questions

Add this skill to your agents