/do — Unified Intent Router

Orientation

Use /do when the user wants something done but doesn't know (or care) which tool handles it. Don't use when: you know the destination — invoke /marshal, /archon, /fleet, or any skill directly.

Commands

Command	Behavior
/do [anything]	Classify intent, route to cheapest capable path
/do status	Show full harness dashboard (/dashboard)
/do continue	Resume most recent active campaign or fleet session
/do --list	Show all skills grouped by category with trigger keywords
/do setup	First-run experience — configure the harness for this project

Protocol

Classification runs top-to-bottom. First match wins. Each tier is cheaper than the next.

Step 0: Skill Registry Check (Cost: ~0 on hit | ~50 tokens on miss)

Before routing, check if new skills have been added since last registration.

1. Count installed Citadel skills (built-in from plugin + custom in project's .claude/skills/)
2. Read registeredSkillCount from .claude/harness.json
3. If counts match: continue to Tier 0. Zero cost.
4. If counts differ (or harness.json doesn't exist yet): a. Read the registeredSkills array from harness.json (default: []) b. Diff skill names against the registered list c. For each unknown skill: read ONLY lines 1-10 of its SKILL.md (frontmatter) d. Extract name and description from frontmatter e. Add the skill to the Tier 2 keyword table for this session using its name and description words as match targets f. Log to the user: "Discovered {N} new skill(s): {names}. Run /do setup to permanently register routing keywords." g. Update registeredSkillCount and registeredSkills array in harness.json

This means:

• 99% of invocations: one number comparison, zero file reads
• New skill dropped in: reads only the new frontmatter, routes immediately
• /do setup does a full registry rebuild with permanent keyword assignment

Tier 0: Pattern Match (Cost: ~0 tokens | Latency: <1ms)

Regex/keyword on raw input. Catches trivial commands:

Pattern	Action
"typecheck" or "type check"	Run the project's typecheck command
"build"	Run the project's build command
"test" or "tests"	Run the project's test command
"status", "dashboard", "what's happening", "what's going on", "show activity"	Show full harness dashboard (/dashboard)
"continue" or "keep going"	Resume active campaign or fleet session
"setup"	Run /do setup first-run experience
"--list" or "list"	Show all available skills
"fix typo in X" or "rename X to Y"	Direct edit (no orchestrator needed)
"commit"	Stage and commit changes
"rollback", "undo phase", "restore checkpoint"	Find active campaign, read latest checkpoint ref, run git stash pop

If matched → execute directly. Done.

Tier 1: Active State Short-Circuit (Cost: ~0 tokens | Latency: <100ms)

Check for active campaigns or fleet sessions that match the input scope:

1. Read .planning/campaigns/ for files with Status: active or status: active in frontmatter
2. Read .planning/fleet/ for session files with status: active or needs-continue
3. Improve campaigns (type: improve): if the active campaign has type: improve in frontmatter, route to /improve {target} --continue where {target} is the campaign's target field. Do NOT route improve campaigns to archon -- improve is its own orchestrator.
4. If input scope matches a non-improve active campaign → /archon continue
5. If fleet session needs continuation → /fleet continue
6. If input mentions a campaign by name → resume it (check type field for routing)
7. If input is "continue" but NO active campaign or fleet session found:
   • Output: "No active campaign or fleet session found. Nothing to continue."
   • If .planning/daemon.json exists with status: "running": the daemon spawned this session but there's no work to do. Update daemon.json: status: "stopped", stopReason: "no-active-work", stoppedAt: "{ISO timestamp}". Delete both triggers if IDs are present. Output: "[daemon] Stopped -- no active campaign found. The work is done."
   • Exit. Do NOT fall through to Tier 2 or 3.

If matched → resume the active work. Done.

Tier 2: Skill Keyword Match (Cost: ~0 tokens | Latency: <10ms)

Match input against installed skill keywords from Citadel's built-in skills and any project-level custom skills in .claude/skills/.

Built-in skill triggers:

Input Contains	Route To
"prd", "requirements", "spec", "plan an app", "design an app"	/prd
"architect", "architecture", "design the system", "file structure", "plan the build"	/architect
"create app", "build app", "build me", "make an app", "new app", "generate app"	/create-app
"add [feature]", "implement [feature]", "add auth", "add payments", "integrate [x]"	/create-app (Tier 5 — feature mode)
"review", "code review"	/review
"test", "generate tests", "write tests"	/test-gen
"document", "docs", "docstring", "readme"	/doc-gen
"refactor", "rename", "extract", "split file"	/refactor
"scaffold", "new module", "new component", "bootstrap"	/scaffold
"create skill", "new skill", "repeated pattern"	/create-skill
"handoff", "session summary"	/session-handoff
"orchestrate", "chain skills", "multi-step"	/marshal
"campaign", "multi-session", "phases"	/archon
"parallel", "simultaneous", "multiple agents", "at the same time", "both ... and"	/fleet --quick
"intake", "process pending", "pipeline"	/autopilot
"setup", "first run", "configure harness"	/setup
"research", "investigate", "look into", "find out"	/research
"experiment", "optimize", "try", "A/B", "measure"	/experiment
"debug", "root cause", "diagnose", "why is", "investigate bug"	/systematic-debugging
"research fleet", "parallel research", "multi-angle research", "compare options"	/research-fleet
"preview", "screenshot", "visual check", "does it render"	/live-preview
"postmortem", "retro", "what broke", "what happened", "debrief"	/postmortem
"design", "style guide", "design manifest", "visual consistency"	/design
"qa", "test the app", "click through", "does it work", "browser test"	/qa
"triage", "open issues", "unlabeled issues", "review pr", "review prs", "investigate issue"	/triage
"watch pr", "watch ci", "monitor pr", "fix ci", "ci failing", "pr failing", "auto-fix", "auto fix pr", "pr is red", "checks failing"	/pr-watch
"dashboard", "what's happening", "what's going on", "show activity", "harness state", "show me status"	/dashboard
"telemetry", "what did this cost", "session cost", "how much did that cost", "how much have I spent", "what hooks fired", "trust level", "show me telemetry", "spending", "session stats", "what telemetry"	/telemetry
"learn", "extract patterns", "learn from that", "save what worked", "patterns from campaign"	/learn
"schedule", "recurring", "every N minutes", "cron", "set a reminder", "run periodically"	/schedule
"merge review", "check merges", "any conflicts", "fleet conflicts", "pending branches", "safe to merge"	/merge-review
"ascii diagram", "ascii art", "box diagram", "architecture diagram", "flow diagram", "draw a diagram", "text diagram", "sequence diagram"	/ascii-diagram
"improve", "improvement loop", "quality loop", "rubric", "score against", "run improvement", "improve citadel"	/improve
"evolve", "sustained improve", "improvement director", "research-driven improve", "multi-cycle improve", "run until done", "improve until ceiling", "keep improving", "hypothesis", "belief model", "scout agents"	/evolve
"organize", "directory structure", "folder structure", "project structure", "file organization", "where should this go", "cleanup directories"	/organize
"houseclean", "house clean", "disk space", "free space", "drive full", "running out of space", "clean up disk", "clean worktrees", "disk audit", "storage audit", "move to another drive", "free up space", "c drive full", "orphaned worktrees"	/houseclean
"daemon", "continuous", "run overnight", "keep running", "24/7", "unattended", "run autonomously", "daemon start", "daemon stop", "daemon status"	/daemon
"map", "index codebase", "codebase map", "structural index", "scan codebase", "map stats", "map query"	/map
"watch", "watch files", "watch changes", "file sentinel", "monitor files", "watch start", "watch stop", "watch scan", "marker comments", "@citadel"	/watch
"infra", "infrastructure", "what databases", "what systems", "docker-compose", "infra audit", "map infrastructure", "what does this connect to"	/infra-audit
"workspace", "multi-repo", "cross-repo", "across repos", "multiple repos", "coordinate repos", "add redis and snowflake", "split into repos"	/workspace

If ONE skill matches with high confidence → invoke it directly. Done. High confidence = evaluator assigns ≥ 0.85 probability to exactly one skill. Below 0.85, or multiple skills above 0.70, fall through to Tier 3. If MULTIPLE skills match → carry the candidate set to Tier 3. Tier 3 disambiguates between candidates only, not from scratch. Tie-break: prefer the candidate with fewer trigger keywords.

Tier 3: LLM Complexity Classifier (Cost: ~500 tokens | Latency: ~1-2s)

When Tiers 0-2 don't resolve, classify across 6 dimensions:

SCOPE: single-file | single-domain | cross-domain | platform-wide
COMPLEXITY: 1 (trivial) | 2 (simple) | 3 (moderate) | 4 (complex) | 5 (campaign)
INTENT: fix | build | create | add | audit | redesign | research | improve | wire | prune
REQUIRES_PERSISTENCE: true | false (multi-session?)
REQUIRES_PARALLEL: true | false (independent sub-tasks?)
REQUIRES_TASTE: true | false (quality judgment beyond tests?)

Routing rules (first match wins):

Condition	Route
INTENT is "create", Complexity >= 3	/create-app
INTENT is "create", Complexity <= 2	/scaffold
INTENT is "add", existing source files present	/create-app (Tier 5 — feature mode)
INTENT is "add", no existing source files	/scaffold
Complexity 1, single skill match	Skill directly
Complexity 1, no skill match	Do it yourself (direct edit)
Complexity 2, single domain	/marshal
Complexity 2-3, known skill domain	Skill, with marshal fallback
Complexity 3, cross-domain	/marshal
Complexity 3-4, requires persistence	/archon
Complexity 4, requires taste/judgment	/archon
Complexity 4-5, requires parallel	/fleet
Complexity 5, platform-wide	/fleet
Confidence < 0.7	/marshal (safe default)

Important: A repeated pattern complaint ("I keep doing X manually", "the agent always makes this mistake") should route to /create-skill. A repeated pattern is a skill waiting to be extracted.

Step 3.5: Proportionality Check

After classification and before execution, verify the response is proportional to the input:

Downgrade triggers (apply in order):

Condition	Action
Input < 20 words AND routed to Archon or Fleet	Downgrade to Marshal. Log: "Input too brief for campaign-level orchestration."
Input mentions a single file AND routed to Fleet	Downgrade to Marshal or skill. Log: "Single-file scope doesn't warrant parallel agents."
Estimated sessions > 5 AND user is Novice trust level	Cap at 3 sessions. Log: "Capping sessions for novice user. Run more to unlock higher budgets."
Routed to Daemon AND user is Novice trust level	Block. Output: "Daemon mode requires familiarity with the harness. Complete a few sessions first."
Estimated cost > $50 AND no explicit budget flag	Confirm with user regardless of trust level.

Upgrade triggers:

Condition	Action
Input complexity >= 4 AND routed to a bare skill	Suggest Marshal. "This looks complex enough for orchestration. Route to /marshal instead?"
Input mentions "overnight" or "continuous" AND routed to Archon	Suggest daemon. "This sounds like continuous work. Want to run it as a daemon?" (skip if Novice)
Input contains 2+ clearly independent tasks AND complexity >= 3	Run Fleet auto-decomposition (see below).

Fleet auto-decomposition — 1/2/3 confirmation prompt:

When 2+ independent tasks detected (non-overlapping scopes, complexity >= 3, not already routed to full Fleet), read consent.fleetSpawn from harness.json:

• auto-allow → route directly to /fleet --quick
• always-ask or null → show prompt: "These look independent — run in parallel? [1=yes 2=always 3=no]"
  • 1: route to --quick, preference unchanged
  • 2: route to --quick, write writeConsent('fleetSpawn', 'auto-allow')
  • 3: run sequentially; if "don't ask again", write always-ask

readConsent/writeConsent are in hooks_src/harness-health-util.js.

Trust level: Read from harness.json trust object. Levels: novice (0-4 sessions), familiar (5-19), trusted (20+ with 2+ campaigns). trust.override takes precedence.

Step 4: After Classification

1. Log routing decision (fire-and-forget): node .citadel/scripts/telemetry-log.cjs --event agent-complete --agent do-router --session routing --status success --meta '{"tier":N,"target":"[skill]","input_chars":M}'

   Use .citadel/scripts/telemetry-log.cjs (the project-local copy). If it doesn't exist, skip logging silently — never block routing on telemetry failure.

2. Announce the routing decision: "Routing to [target] because [one-sentence reason]"

3. Invoke the target skill or orchestrator

4. If the target fails or the user says "wrong tool", try the next tier up. If the target is already Tier 3 (marshal fails or user explicitly escalates from a failed marshal attempt): re-route to /archon with the original input as context.

/do --list

Output a grouped skill list drawn from the system reminder's available skills. Group by category (Orchestration, App Creation, Code Quality, Research & Debugging, GitHub & CI, Infrastructure, Monitoring, Utilities, Observability). For each skill, show /name — one-line description. Include a footer: "Direct invocation (/skill-name) always bypasses the router."

Fringe Cases

• .planning/ does not exist: The router works without .planning/. Tiers 0, 2, and 3 are fully independent of it. Tier 1 (active-state short-circuit) reads .planning/campaigns/ and .planning/fleet/ — if those directories are absent, skip Tier 1 gracefully and fall through to Tier 2. Never crash on a missing .planning/ directory.
• harness.json missing: Skip the Skill Registry Check and proceed directly to Tier 0. Announce discovered skills from the filesystem if counts can be read, otherwise route from built-in keywords.
• Multiple skills match at Tier 2: Carry candidates to Tier 3 per Tier 2 disambiguation rule above.
• User input is empty or whitespace: Respond with the --list output and a prompt to provide a direction.
• Routed skill not found: Report "Skill not found" and fall back to Marshal as the safe default.

Contextual Gates

Disclosure: "Routing to [skill]. See that skill's contextual gates for reversibility." Reversibility: depends on routed skill — check the routed skill's reversibility Trust gates:

• Any: routing and dispatch; inherits trust gates from the routed skill.

Quality Gates

• Tier 0-2 must resolve in under 1 second
• Tier 3 classification must be transparent (announce reasoning)
• Never route a trivial task (complexity 1) to Archon or Fleet
• Never route a multi-session task to a bare skill
• If routing fails, default to Marshal (safe middle ground)

Exit Protocol

After routing and execution complete:

• If the routed skill/orchestrator produces a HANDOFF, relay it to the user
• If the task was trivial (Tier 0), just show the result
• Do not add overhead to simple tasks
• Telemetry is fire-and-forget — never surface telemetry errors to the user