Gladiator Plugin
Continuous learning. Observes tool failures and prompts reflection to evolve rules, hooks, and skills.
Hooks
| Hook |
When |
Action |
| PostToolUseFailure(Bash/Edit) |
After tool failure |
Nudges you to call gladiator_observe() — you must call it, not automatic |
| Stop |
Session ending |
Prompts gladiator_reflect() if >= 10 unprocessed observations |
Commands
| Command |
Description |
/review-gladiator [topic] |
Batch learn from accumulated observations and session history |
Workflows
Observe (manual — hooks only nudge)
After a tool failure, the hook nudges you. YOU must call gladiator_observe() — it is NOT automatic. After fixing an error, record what worked:
gladiator_observe(
summary: "<what failed and how it was fixed>",
context: {error, tool, before, after},
tags: ["error", "<tool_name>"]
)
Reflect
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
Batch Review (/review-gladiator)
- Review accumulated observations across sessions
gladiator_reflect() to cluster all observations- Present recommendations to user
- Apply updates to existing rules/skills/hooks
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
1---2name: claude-gladiator3description: Continuous learning — INVOKE after sessions with many errors or corrections to record patterns and improve rules, hooks, and skills4---56# Gladiator Plugin78Continuous learning. Observes tool failures and prompts reflection to evolve rules, hooks, and skills.910## Hooks1112| Hook | When | Action |13|------|------|--------|14| **PostToolUseFailure(Bash/Edit)** | After tool failure | Nudges you to call `gladiator_observe()` — you must call it, not automatic |15| **Stop** | Session ending | Prompts `gladiator_reflect()` if >= 10 unprocessed observations |1617## Commands1819| Command | Description |20|---------|-------------|21| `/review-gladiator [topic]` | Batch learn from accumulated observations and session history |2223## Workflows2425### Observe (manual — hooks only nudge)2627After a tool failure, the hook nudges you. YOU must call `gladiator_observe()` — it is NOT automatic. After fixing an error, record what worked:2829```30gladiator_observe(31 summary: "<what failed and how it was fixed>",32 context: {error, tool, before, after},33 tags: ["error", "<tool_name>"]34)35```3637### Reflect38391. `gladiator_reflect()` — cluster observations into recommendations402. For each recommendation: read the existing artifact (if overlap detected)413. Propose UPDATE to existing artifact, not a new duplicate424. Present to user with reasoning435. Apply changes one at a time after approval4445### Batch Review (/review-gladiator)46471. Review accumulated observations across sessions482. `gladiator_reflect()` to cluster all observations493. Present recommendations to user504. Apply updates to existing rules/skills/hooks5152## Observation Templates5354| Situation | Call |55|-----------|------|56| Tool failure (auto) | `gladiator_observe(summary, context={error, tool, before, after}, tags=["error", tool])` |57| User correction | `gladiator_observe(summary, context={before, after}, tags=["correction"])` |58| Convention found | `gladiator_observe(summary, tags=["convention", "domain"])` |59| Decision made | `gladiator_observe(summary, tags=["architecture", "decision"])` |6061## Requires6263```64claude mcp add gladiator -- npx claude-gladiator-mcp65```
vvkmnn