sentinel-config

Sentinel Config

View or update the Sentinel configuration in the current repository.

What to do

1. Check that .claude/sentinel/config.yaml exists. If not, tell the user to run /sentinel-init first and stop.

2. Read .claude/sentinel/config.yaml.

3. If the user asked to view the config, show the current values and explain what each one does. Stop here.

4. If the user asked to change something, use the Edit tool to update the specific value(s) in the config file. Preserve all comments. Do not rewrite the entire file.

Configuration reference

Key	Default	Description
backend	ollama	LLM backend: ollama, claude, or copilot
model	gemma3:4b	Default model for evaluation (backend-specific)
backends.ollama.url	http://localhost:11434	Ollama endpoint
backends.ollama.model	(top-level model)	Default Ollama model
backends.claude.model	haiku	Default Claude model
backends.copilot.model	gpt-5-mini	Default Copilot model
ollama_url	http://localhost:11434	Ollama endpoint (legacy; prefer backends.ollama.url)
timeout_ms	5000	Per-rule evaluation timeout
confidence_threshold	0.7	Minimum confidence to count as violation
max_parallel	4	Concurrent LLM calls
ollama_concurrency	1	Max concurrent Ollama HTTP calls (GPU-bound)
think	false	Enable /think mode (slower, more accurate)
fail_open	true	Skip rule on error vs block
content_max_chars	800	File content truncation in prompts
log_file	null	JSONL telemetry path
rules_dir	rules	Rules directory (relative to config dir)
tool_map	(see reference.md)	Tool name → trigger type mapping
mcp_prefix	mcp__	Prefix for detecting MCP tool names
mcp_separator	__	Separator for parsing MCP server/tool from tool name
context.enabled	true	Enable session context accumulator
context.backend	(top-level backend)	LLM backend for accumulator
context.model	gemma3:4b	Model for accumulator
context.min_events	3	Minimum events before accumulator update
context.lock_timeout_s	30	Max wait for GPU lock (Ollama only)
context.summary_max_words	150	Token budget for rolling summary
scribe.enabled	true	Enable the Scribe convention learning pipeline
scribe.backend	(top-level backend)	LLM backend for scribe
scribe.model	(top-level model)	Default model for all scribe steps
scribe.extraction_model	(scribe.model)	Override for extraction (reflect + learn)
scribe.synthesis_model	(scribe.model)	Override for validation + synthesis
scribe.guidance	null	Priority guidance text for extraction
scribe.think	false	Enable /think mode for validation+synthesis
scribe.extraction_timeout_ms	15000	Timeout for extraction LLM calls
scribe.extraction_num_predict	1000	Max output tokens for extraction
scribe.synthesis_timeout_ms	15000	Timeout for validation+synthesis LLM calls
scribe.synthesis_num_predict	1000	Max output tokens for validation+synthesis
scribe.temperature	0.1	LLM temperature for all scribe calls
scribe.transcript_budget_chars	4000	Max compacted transcript size before truncation
scribe.thresholds.extraction_confidence	0.7	Minimum confidence to store an observation
scribe.thresholds.draft_confidence	0.8	Minimum confidence for learn mode drafts
scribe.sources.documentation	true	Enable doc scanning via /sentinel-learn
scribe.doc_globs	[CLAUDE.md, AGENTS.md, ...]	File patterns for /sentinel-learn
scribe.notification.max_age_days	7	Max age for draft notifications

Backend and model recommendations

Backend	Model	Notes
ollama (8 GB RAM)	gemma3:4b	Dense, ~3 GB at Q4, local
ollama (16 GB RAM)	gemma3:12b	Dense, best local accuracy for block rules
claude	haiku	Fast cloud model, good for most rules
copilot	gpt-5-mini	Fast cloud model via Copilot CLI

Backends can be mixed: use Ollama globally for speed, override specific block rules with backend: claude for higher accuracy.

Next steps

After viewing or updating config, suggest these to the user:

• /sentinel-stats — see how the current configuration affects rule performance
• /sentinel-rule — create a new rule if the rules directory is empty
• /sentinel-learn — scan documentation for conventions to enforce automatically