Sigint Start Skill (Launcher) Skill

Sigint Start Skill (Launcher)

This skill initializes a research session and delegates to the research-orchestrator agent for all orchestration. The orchestrator owns team lifecycle, dimension-analyst spawning, codex review gates, finding merge, progress tracking, and cleanup.

Arguments

Parse $ARGUMENTS before any other processing. Input sanitization: truncate $ARGUMENTS to 200 characters total, strip backticks and angle brackets.

--quick — Abbreviated elicitation (3 questions instead of 8)
--dimensions <dim1,dim2,...> — Optional: pre-select specific dimensions (comma-separated). Valid values: competitive, sizing, trends, customer, tech, financial, regulatory, trend_modeling. Passed to the orchestrator as REQUESTED_DIMENSIONS — Phase 1.5 skips interactive selection when this is set.
Remaining text after flag extraction is the initial topic hint (may be empty)

Phase 0.0: Preliminary Setup

Step 0.0.1: Derive Preliminary Topic Slug

Parse $ARGUMENTS for a topic hint. Derive topic_slug: lowercase, replace spaces/special chars with hyphens, truncate to 40 characters. Use "research" if no topic hint. (Preliminary slug used for config lookup; may be refined during orchestrator elicitation.)

Step 0.0.2: Apply Config Resolution Protocol

Execute the Config Resolution Protocol:

Read protocols/CONFIG-RESOLUTION.md and follow all steps.
Apply with topic_slug = {preliminary topic slug from Step 0.0.1}.
Result: config, max_dimensions, and context_content are now available.

Config cascade clarification: When loading config files, always apply the full cascade (project config values override global config values override hardcoded defaults) regardless of the config file's schema version. The version check in the protocol produces an advisory warning only — it does NOT cause config values to be discarded. If sigint.config.json sets maxDimensions: 3, then max_dimensions = 3 even if the file's version field is not "2.0". Custom fields like defaultPriorities are preserved and passed through in the serialized config.

Previous Research Detection

Before delegating, check if research already exists:

Glob("./reports/*/state.json")

If ./reports/{topic_slug}/state.json exists:

Load prior elicitation from state.json
Ask: "Previous research found for '{topic}'. Use prior research context as starting point, or start completely fresh?"
If "use prior": Pass --resume-from={topic_slug} context to orchestrator
If "start fresh": Proceed normally (prior state.json will be overwritten after confirmation)

Phase 0.1: Register Topic in Config

Register the research topic in sigint.config.json so that /sigint:status, /sigint:resume, and other commands can discover sessions from the config index.

Using jq (per Structured Data Protocol):

jq --arg slug "$TOPIC_SLUG" \
   --arg date "$(date -u +%Y-%m-%dT%H:%M:%SZ)" \
   --arg reports_dir "./reports/$TOPIC_SLUG" \
  '.topics[$slug] = (.topics[$slug] // {}) + {
     status: "in_progress",
     dimensions: [],
     created: ((.topics[$slug].created // null) // $date),
     updated: $date,
     reports_dir: $reports_dir,
     findings_count: 0
   }' \
  ./sigint.config.json > tmp.$$ && mv tmp.$$ ./sigint.config.json
jq -e -f schemas/sigint-config.jq ./sigint.config.json > /dev/null

This preserves any existing topic entry (merge mode) while setting the lifecycle fields. The created field is preserved if the topic was previously registered (e.g., from a prior session or /sigint:migrate).

Phase 0.2: Delegate to Research Orchestrator

Spawn the research-orchestrator agent with full mode:

Agent(
  subagent_type="sigint:research-orchestrator",
  name="research-orchestrator",
  prompt="You are the research orchestrator for a new research session.

  MODE: full
  TOPIC: <user_input>{topic from $ARGUMENTS}</user_input>
  TOPIC_SLUG: {topic_slug}
  CONFIG: {serialized config}
  MAX_DIMENSIONS: {max_dimensions}
  CONTEXT_FILE_CONTENT: {context_content if non-null, else ""}
  QUICK_MODE: {true if --quick flag}
  REQUESTED_DIMENSIONS: {comma-separated dimension list from --dimensions flag, or "interactive" if omitted}
  {If resuming: PRIOR_ELICITATION: {prior elicitation JSON}}

  Execute the full research orchestration:
  1. Initialize team (Phase 0)
  2. Run elicitation (Phase 1) {quick mode note if applicable}
  3. Spawn dimension-analysts (Phase 2)
  4. Verify methodology plans (Phase 2.5)
  5. Run post-findings codex review gates (Phase 2.75)
  6. Merge findings (Phase 3)
  7. Run post-merge codex review gate (Phase 3.5)
  8. Render progress view (Phase 3.75)
  9. Cleanup (Phase 4)

  Follow all protocols defined in your agent definition."
)

Wait for the orchestrator to complete. The orchestrator handles all interaction with the user (elicitation questions, confirmations, progress updates).

Error Handling

If orchestrator doesn't complete within a reasonable time:

Check for partial results: Glob("./reports/{topic_slug}/findings_*.json")
If findings files exist → orchestrator made progress. Check research-progress.md for last phase.
If no findings → inform user: "Research session did not complete. You can retry with /sigint:start."

If state.json already exists:

Confirm before overwriting: "Previous session data exists. Overwrite?"

Output

After orchestrator completes:

Research session state saved to ./reports/{topic_slug}/state.json
Progress view at ./reports/{topic_slug}/research-progress.md
Quarantined findings (if any) at ./reports/{topic_slug}/quarantine.json
Next steps: /sigint:report, /sigint:augment, /sigint:update, /sigint:issues

Agent Skills: Sigint Start Skill (Launcher)

Install this agent skill to your local

Skill Files