claude-gladiator

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

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

Batch Review (/review-gladiator)

1. Review accumulated observations across sessions
2. gladiator_reflect() to cluster all observations
3. Present recommendations to user
4. 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