skill-router

Name: skill-router
Author: Rune-kit

80stars21forks

How to install skill-router?

rune-kit/rune/skill-router

$npx -y skills add Rune-kit/rune --skill skill-router

Installs into the current project.

Summary

Meta-enforcement layer that routes EVERY agent action through the correct skill. MUST check this routing table before ANY response involving code, files, or technical decisions. Default: route to rune:cook for code tasks. Prevents rationalization, enforces check-before-act discip

Files · 1

View on GitHub

SKILL.md

1	## Live Routing Context
2
3	Routing overrides (if available): !`cat .rune/metrics/routing-overrides.json 2>/dev/null \|\| echo "No adaptive routing rules active."`
4
5	Recent skill usage: !`cat .rune/metrics/skills.json 2>/dev/null \| head -20 \|\| echo "No metrics collected yet."`
6
7	# skill-router
8
9	## Purpose
10
11	The missing enforcement layer for Rune. While individual skills have HARD-GATEs and constraints, nothing forces the agent to check for the right skill before acting. `skill-router` fixes this by intercepting every user request and routing it through the correct skill(s) before any code is written, any file is read, or any clarifying question is asked.
12
13	This is L0 — it sits above L1 orchestrators. It doesn't do work itself; it ensures the right skill does the work.
14
15	## Triggers
16
17	- ALWAYS — This skill is conceptually active on every user message
18	- Loaded via system prompt or plugin description, not invoked manually
19	- The agent MUST internalize this routing table and apply it before every response
20
21	## Calls (outbound connections)
22
23	- Any skill (L1-L3): routes to the correct skill based on intent detection
24
25	## Called By (inbound connections)
26
27	- None — this is the entry point. Nothing calls skill-router; it IS the first check.
28
29	## Workflow
30
31	### Step 0 — Check Routing Overrides (H3 Adaptive Routing)
32
33	Before standard routing, check if adaptive routing rules exist:
34
35	1. Use `Read` on `.rune/metrics/routing-overrides.json`
36	2. If the file exists and has active rules, scan each rule's `condition` against the current user intent
37	3. If a rule matches:
38	- Apply the override action (e.g., "route to problem-solver before debug")
39	- Log: "Adaptive routing: applying rule [id] — [action]"
40	4. If no file exists or no rules match, proceed to standard routing (Step 1)
41
42	Override constraints:
43	- Overrides MUST NOT bypass layer discipline (L3 cannot call L1)
44	- Overrides MUST NOT skip quality gates (sentinel, preflight, verification)
45	- Overrides MUST NOT route to non-existent skills
46	- If an override seems wrong, announce it and let user decide to keep or disable
47
48	Model hint support (Adaptive Model Re-balancing):
49	- Override entries may include `"model_hint": "opus"` — this signals that a skill previously failed at sonnet-level and needed opus reasoning depth
50	- When a model_hint is present, announce: "Adaptive routing: this skill previously required opus-level reasoning for [context]. Escalating model."
51	- Model hints are written by cook Phase 8 when debug-fix loops hit max retries on the same error pattern
52	- Model hints do NOT override explicit user model preferences
53
54	### Context Efficiency (Trigger-Table Pattern)
55
56	Skill-router's routing table above IS the trigger table — it maps keywords to skill paths without loading any skill content. Skills are loaded on-demand via the Skill tool only when routed. This keeps baseline context usage minimal.
57
58	Rules for context efficiency:
59	- NEVER read a SKILL.md to decide routing — use the routing table keywords
60	- NEVER load multiple skills speculatively — route to ONE, let it chain if needed
61	- Skill content is loaded by the Skill tool, not by skill-router reading files
62
63	### Step 0.25 — Request Classifier (Fast-Path Filter)
64
65	Before intent classification, categorize the request into one of 5 types. This determines the enforcement level — how strictly routing must be followed.
66
67	\| Request Type \| Keywords / Signals \| Enforcement \| Action \|
68	\|---\|---\|---\|---\|
69	\| `CODE_CHANGE` \| "build", "implement", "add", "create", "fix", "refactor", "update code" \| FULL \| cook mandatory, no exceptions \|
70	\| `QUESTION` \| "what is", "how does", "explain", "why" \| LITE \| Check if a skill has domain knowledge first; answer directly if no skill matches \|
71	\| `DEBUG_REQUEST` \| "error", "bug", "not working", "broken", "crash", "fails" \| FULL \| debug skill mandatory \|
72	\| `REVIEW_REQUEST` \| "review", "check", "audit", "look at this code" \| FULL \| review skill mandatory \|
73	\| `EXPLORE` \| "find", "search", "where is", "show me", "list" \| LITE \| scout if codebase-related; answer directly if general \|
74
75	Enforcement levels:
76	- FULL → MUST route through a skill. Writing code without skill invocation = protocol violation.
77	- LITE → SHOULD check if a skill applies. Can answer directly if no skill matches and the response involves no code changes.
78
79	Escape hatch: If request is clearly trivial (< 5 LOC change, single-line fix, user says "just do it"), classify as CODE_CHANGE but cook activates Fast Mode automatically.
80
81	### Step 0.3

People who liked this also viewed

6 picks

Type

skill-router

80stars21forks

View on GitHub

How to install skill-router?

rune-kit/rune/skill-router

$npx -y skills add Rune-kit/rune --skill skill-router

Installs into the current project.

Summary

Files · 1

View on GitHub

SKILL.md

1	## Live Routing Context
2
3	Routing overrides (if available): !`cat .rune/metrics/routing-overrides.json 2>/dev/null \|\| echo "No adaptive routing rules active."`
4
5	Recent skill usage: !`cat .rune/metrics/skills.json 2>/dev/null \| head -20 \|\| echo "No metrics collected yet."`
6
7	# skill-router
8
9	## Purpose
10
11	The missing enforcement layer for Rune. While individual skills have HARD-GATEs and constraints, nothing forces the agent to check for the right skill before acting. `skill-router` fixes this by intercepting every user request and routing it through the correct skill(s) before any code is written, any file is read, or any clarifying question is asked.
12
13	This is L0 — it sits above L1 orchestrators. It doesn't do work itself; it ensures the right skill does the work.
14
15	## Triggers
16
17	- ALWAYS — This skill is conceptually active on every user message
18	- Loaded via system prompt or plugin description, not invoked manually
19	- The agent MUST internalize this routing table and apply it before every response
20
21	## Calls (outbound connections)
22
23	- Any skill (L1-L3): routes to the correct skill based on intent detection
24
25	## Called By (inbound connections)
26
27	- None — this is the entry point. Nothing calls skill-router; it IS the first check.
28
29	## Workflow
30
31	### Step 0 — Check Routing Overrides (H3 Adaptive Routing)
32
33	Before standard routing, check if adaptive routing rules exist:
34
35	1. Use `Read` on `.rune/metrics/routing-overrides.json`
36	2. If the file exists and has active rules, scan each rule's `condition` against the current user intent
37	3. If a rule matches:
38	- Apply the override action (e.g., "route to problem-solver before debug")
39	- Log: "Adaptive routing: applying rule [id] — [action]"
40	4. If no file exists or no rules match, proceed to standard routing (Step 1)
41
42	Override constraints:
43	- Overrides MUST NOT bypass layer discipline (L3 cannot call L1)
44	- Overrides MUST NOT skip quality gates (sentinel, preflight, verification)
45	- Overrides MUST NOT route to non-existent skills
46	- If an override seems wrong, announce it and let user decide to keep or disable
47
48	Model hint support (Adaptive Model Re-balancing):
49	- Override entries may include `"model_hint": "opus"` — this signals that a skill previously failed at sonnet-level and needed opus reasoning depth
50	- When a model_hint is present, announce: "Adaptive routing: this skill previously required opus-level reasoning for [context]. Escalating model."
51	- Model hints are written by cook Phase 8 when debug-fix loops hit max retries on the same error pattern
52	- Model hints do NOT override explicit user model preferences
53
54	### Context Efficiency (Trigger-Table Pattern)
55
56	Skill-router's routing table above IS the trigger table — it maps keywords to skill paths without loading any skill content. Skills are loaded on-demand via the Skill tool only when routed. This keeps baseline context usage minimal.
57
58	Rules for context efficiency:
59	- NEVER read a SKILL.md to decide routing — use the routing table keywords
60	- NEVER load multiple skills speculatively — route to ONE, let it chain if needed
61	- Skill content is loaded by the Skill tool, not by skill-router reading files
62
63	### Step 0.25 — Request Classifier (Fast-Path Filter)
64
65	Before intent classification, categorize the request into one of 5 types. This determines the enforcement level — how strictly routing must be followed.
66
67	\| Request Type \| Keywords / Signals \| Enforcement \| Action \|
68	\|---\|---\|---\|---\|
69	\| `CODE_CHANGE` \| "build", "implement", "add", "create", "fix", "refactor", "update code" \| FULL \| cook mandatory, no exceptions \|
70	\| `QUESTION` \| "what is", "how does", "explain", "why" \| LITE \| Check if a skill has domain knowledge first; answer directly if no skill matches \|
71	\| `DEBUG_REQUEST` \| "error", "bug", "not working", "broken", "crash", "fails" \| FULL \| debug skill mandatory \|
72	\| `REVIEW_REQUEST` \| "review", "check", "audit", "look at this code" \| FULL \| review skill mandatory \|
73	\| `EXPLORE` \| "find", "search", "where is", "show me", "list" \| LITE \| scout if codebase-related; answer directly if general \|
74
75	Enforcement levels:
76	- FULL → MUST route through a skill. Writing code without skill invocation = protocol violation.
77	- LITE → SHOULD check if a skill applies. Can answer directly if no skill matches and the response involves no code changes.
78
79	Escape hatch: If request is clearly trivial (< 5 LOC change, single-line fix, user says "just do it"), classify as CODE_CHANGE but cook activates Fast Mode automatically.
80
81	### Step 0.3

People who liked this also viewed

6 picks

Type